Download Colloidal synthesis of metal oxide nanocrystals and thin films Fredrik Söderlind

Linköping Studies in Science and Technology
Dissertation No. 1182
Colloidal synthesis of metal oxide
nanocrystals and thin films
Fredrik Söderlind
Division of Chemistry
Department of Physics, Chemistry and Biology (IFM)
Linköping University, SE-581 83 Linköping, Sweden
Linköping 2008
Cover images
GdFeO3: the structure, a nanocrystal, and a thin film.
Copyright © 2008 Fredrik Söderlind, unless otherwise noted.
ISBN 978-91-7393-899-0
ISSN 0345-7524
Printed by LiU-Tryck, Linköping, Sweden.
“But if you could just see the beauty
These things I could never describe
These pleasures a wayward distraction
This is my one lucky prize”
Joy Division, 1980.
“Mistakes are the portals of discovery”
James Joyce
ABSTRACT
A main driving force behind the recent years’ immense interest in
nanoscience and nanotechnology is the possibility of achieving new material
properties and functionalities within, e.g., material physics, biomedicine,
sensor technology, chemical catalysis, energy storing systems, and so on.
New (theoretical) possibilities represent, in turn, a challenging task for
chemists and physicists. An important feature of the present nanoscience
surge is its strongly interdisciplinary character, which is reflected in the
present work.
In this thesis, nanocrystals and thin films of magnetic and
ferroelectric metal oxides, e.g. RE2O3 (RE = Y, Gd, Dy), GdFeO3,
Gd3Fe5O12, Na0.5K0.5NbO3, have been prepared by colloidal and sol-gel
methods. The sizes of the nanocrystals were in the range 3-15 nm and
different carboxylic acids, e.g. oleic or citric acid, were chemisorbed onto the
surface of the nanoparticles. From FT-IR measurements it is concluded that
the bonding to the surface takes place via the carboxylate group in a
bidentate or bridging fashion, with some preference for the latter
coordination mode. The magnetic properties of nanocrystalline Gd2O3 and
GdFeO3 were measured, both with respect to magnetic resonance relaxivity
and magnetic susceptibility. Both types of materials exhibit promising
relaxivity properties, and may have the potential for use as positive contrast
enhancing agents in magnetic resonance imaging (MRI). The nanocrystalline
samples were also characterized by transmission electron microscopy (TEM),
x-ray photoelectron spectroscopy (XPS), and quantum chemical calculations.
Thin films of Na0.5K0.5NbO3, GdFeO3 and Gd3Fe5O12 were prepared
by sol-gel methods and characterized by x-ray powder diffraction (XRPD)
and scanning electron microscopy (SEM). Under appropriate synthesis
conditions, rather pure phase materials could be obtained with grain sizes
ranging from 50 to 300 nm. Magnetic measurements in the temperature
range 2-350 K indicated that the magnetization of the perovskite phase
GdFeO3 can be described as the sum of two contributing terms. One term
(mainly) due to the spontaneous magnetic ordering of the iron containing
sublattice, and the other a susceptibility term, attributable to the
paramagnetic gadolinium sublattice. The two terms yield the relationship
M(T) = M0(T) + χ(T) ⋅ H for the magnetization. The garnet phase
Gd3Fe5O12 is ferrimagnetic and showed a compensation temperature
Tcomp ≈ 295 K.
i
ii
SAMMANFATTNING
Nanovetenskap och nanoteknologi är områden som idag får mycket
uppmärksamhet vad gäller forskning och utveckling. Utmaningen för
forskare är att kunna framställa material i nanometerstorlek (i form av t.ex.
nanokristaller eller tunna filmer) och utnyttja de nya egenskaper dessa
material kan få, då egenskaperna kan skilja sig väsentligt från dem man
möter på såväl atomär som makroskopisk nivå. Nanokristaller har funnit
användningsområden inom såväl medicin som kemi, liksom inom sensor-,
datalagrings- och solcellsmaterial. Tunna filmer, som oftast är en
ytbeläggning på upp till 1000 nanometer, används på skärverktyg för att göra
dem mer slitstarka, men även för t.ex. sensorer, hårddiskar, tonade
fönsterglas m.m.
Detta arbete beskriver ett antal synteser av nanokristallina
oxidmaterial av sällsynta jordartsmetaller såsom Gd2O3, Dy2O3 och Y2O3,
men även av järninnehållande föreningar som GdFeO3. Ett av syftena med
denna forskning är att finna material, som i nanokristallin form kan
användas som positivt kontrastmedel i magnetresonanstomografi. Idag
används vid vissa medicinska undersökningar nanokristaller av bl.a. magnetit,
Fe3O4, som ger negativ kontrast, och gadoliniuminnehållande komplex (bl.a.
Gd-DTPA) för positiv kontrast. Gadoliniumjonen Gd3+ är paramagnetisk
med sju oparade 4f-elektroner och påverkar relaxationshastigheten hos de
exciterade väteatomernas kärnspinn (väteatomer finns som bekant i vatten,
och vatten finns i stor mängd i människokroppen) och ger därmed upphov
till positiv (förstärkt) kontrastverkan i den tomografiska bilden.
Nanokristaller innehållande Gd3+ såsom Gd2O3 eller GdFeO3, kan tänkas
förbättra kontrasten ytterligare, då magnetfältet runt varje enskild partikel
förväntas vara starkare än det som omger en enskild komplexbunden Gd3+jon.
Denna studie har så här långt visat att små nanokristaller (< 5 nm) av
både Gd2O3 och GdFeO3 av allt att döma förkortar relaxationstiden
påtagligt vid magnetresonansmätningar. Storlek och form på partiklarna har
studerats med transmissionselektronmikroskopi (TEM) och interaktionen av
organiska molekyler, bl.a. oljesyra, citronsyra m.fl., med kristallytan har
studerats med infrarödspektroskopi (FT-IR) samt röntgenfotoelektronspektroskopi (XPS). FT-IR, tillsammans med teoretiska (kvantkemiska)
beräkningar, har visat att karboxylgruppens två syreatomer binder in till
partikelytan antingen med båda syreatomerna till samma metallatom
iii
(bidentate), eller till varsin metallatom (bridge). För att möjliggöra
användning in vivo är det nödvändigt att partiklarna både är tillräckligt stabila
och kan dispergeras i vattenlösning. De måste dessutom vara tillräckligt små
för att utan problem kunna utsöndras via njurarna, då gadolinium,
åtminstone högre doser, anses toxiskt.
Tunna filmer av det piezoelektriska materialet Na0.5K0.5NbO3 (NKN),
och de magnetiskt intressanta materialen GdFeO3 och Gd3Fe5O12 har
framställts med olika våtkemiska så kallade sol-gelmetoder. Filmerna har
karaktäriserats med röntgendiffraktion (XRD) och svepelektronmikroskopi
(SEM). XRD visade att filmerna är polykristallina och i stort sett fria från
föroreningsfaser. Med SEM studerades filmernas kristallstorlek, ytstruktur
och tjocklek. Kristalliternas storlek, som styrs av den temperatur och tid vid
vilken filmerna sintras, varierade mellan 50 och 300 nm, medan filmskiktets
tjocklek beror av bland annat utgångslösningens viskositet samt hur många
lager lösning som appliceras. De flesta filmer hade en tjocklek runt 300400 nm. Vanligtvis värmebehandlas filmerna vid 600-900 °C. De två
filmerna innehållande Gd och Fe studerades vidare med spektroskopiska
metoder såsom FT-IR och XPS för att ytterligare fastställa filmernas
kemiska beskaffenhet. De magnetiska egenskaperna uppmättes i
temperaturområdet 2-350 K. Magnetiseringen av GdFeO3 kan beskrivas
som summan av två bidragande termer. Den ena härrör från ett spontant
inducerat magnetiskt moment hos järnatomerna, den andra från de
paramagnetiska gadoliniumatomerna. Granatfasen Gd3Fe5O12 uppvisade ett
ferrimagnetiskt beteende med en kompensationtemperatur Tcomp ≈ 295 K.
iv
LIST OF PAPERS
I.
Synthesis and characterisation of Gd2O3 nanocrystals
functionalised by organic acids.
F. Söderlind, H. Pedersen, R.M. Petoral Jr., P.O. Käll, K. Uvdal.
J. Colloid Interface Sci. 288 (2005) 140-148.
II.
Surface interactions between Y2O3 nanocrystals and organic
molecules - an experimental and quantum chemical study.
H. Pedersen, F. Söderlind, R. M. Petoral Jr., K. Uvdal, P.O. Käll,
L. Ojamäe.
Surface Science 592 (2005) 124-140.
III.
Polyethylene glycol-covered ultra-small Gd2O3 nanoparticles
for positive contrast at 1.5 T magnetic resonance clinical
scanning.
M.A. Fortin, R.M. Petoral Jr., F. Söderlind, A. Klasson,
M. Engström, T. Veres, P.O. Käll, K. Uvdal.
Nanotechnology 18 (2007) 395501.
IV.
Colloidal synthesis and characterization of ultrasmall
perovskite GdFeO3 nanocrystals.
F. Söderlind, M.A. Fortin, R.M. Petoral Jr., A. Klasson, T. Veres,
M. Engström, K. Uvdal, P.O. Käll.
Nanotechnology 19 (2008) 085608.
V.
Sol-gel synthesis and characterization of Na0.5K0.5NbO3
(NKN) thin films.
F. Söderlind, P.O. Käll, U. Helmersson.
J. Crystal Growth 281 (2005) 468-474.
VI.
Sol-gel synthesis and characterization of polycrystalline
GdFeO3 and Gd3Fe5O12 thin films.
F. Söderlind, L. Selegård, P. Nordblad, K. Uvdal, P.O. Käll.
Submitted.
v
Further publications (not included in this thesis)
VII.
Synthesis, structure determination and X-ray photoelectron spectroscopy
characterisation of a polymeric silver(I) nicotinic acid complex, H[Ag(py3-CO2)2], P.O. Käll, J. Grins, M. Fahlman, F. Söderlind, Polyhedron, 20 (2001)
2747-2753.
VIII. Low temperature growth and characterization of (Na,K)NbOx thin films,
V.M. Kugler, F. Söderlind, D. Music, U. Helmersson, J. Andreasson, T. Lindbäck,
J. Crystal Growth 254 (2003) 400-404.
IX.
Microstructure/dielectric properties relationship of low temperature
synthesized (Na,K)NbOx thin films, V.M. Kugler, F. Söderlind, D. Music,
U. Helmersson, J. Andreasson, T. Lindbäck, J. Crystal Growth 262 (2004) 322-326.
X.
Positive MRI contrast enhancement in THP-1 cells with Gd2O3
nanoparticles, A. Klasson, M. Ahrén, E. Hellqvist, F. Söderlind, A. Rosén,
P.O. Käll, K. Uvdal, M. Engström, accepted by Contrast Media Mol. Imaging.
XI.
Synthesis and characterisation of Tb3+ doped Gd2O3 nanocrystals, a
bifunctional material with fluorescent labelling and MRI contrast agent
properties, R.M. Petoral Jr., F. Söderlind, A. Klasson, A. Suska, M.A. Fortin,
P.O. Käll, M. Engström, K. Uvdal, submitted to J. Phys. Chem.
XII.
ZnO nanoparticles or ZnO films, a comparison of the gas sensing
capabilities, J. Eriksson, V. Khranovskyy, F. Söderlind, P.O. Käll, R. Yakimova,
A. Lloyd Spetz, submitted to J. Nanoparticle Res.
XIII. Wet synthesis of dichroic “butterfly wing”- shaped gold nanoparticles,
R. Becker, F. Söderlind, A. Herland, B. Liedberg, P.O. Käll, submitted to Nano Lett..
Conference papers
XIV. Synthesis of gadolinium oxide nanoparticles as a contrast agent in MRI,
M.A. Fortin, R.M. Petoral Jr, F. Söderlind, P.O. Käll, M. Engström, K. Uvdal,
Trends in Nanotechnology Grenoble, Schweiz 2006.
XV.
A comparison between the use of Pd- and Au-nanoparticles as sensing
layers in a field effect NOx-sensitive sensor, K. Buchholt, E. Ieva, L. Torsi,
N. Cioffi, L. Colaianni, F. Söderlind, P.O. Käll, A. Lloyd Spetz, 2nd International
Conference on Sensing Technology, ICST, November 26-27, 2007, Palmerston North,
New Zealand, pp. 87-92.
vi
TABLE OF CONTENTS
ABSTRACT
SAMMANFATTNING
LIST OF PAPERS
TABLE OF CONTENTS
ABBREVIATIONS
1. INTRODUCTION
1.1 Materials Chemistry
1.2 Nanoparticles
1.3 Thin Films
1.4 Crystal Structures
1.4.1 Sesquioxides
1.4.2 Perovskites
1.4.3 Garnets
1.4.4 Phase Diagram of the System Fe2O3-Gd2O3
1.5 Material Properties
1.5.1 Electrical Polarization
1.5.2 Piezo- and Ferroelectricity
1.5.3 Magnetic Properties
2. SYNTHESIS METHODS
2.1 Colloidal Synthesis Methods and Nanochemistry
2.2 Electrochemical Methods
2.3 Combustion Methods
2.4 Sol-Gel Methods
2.4.1 Alkoxide Method
2.4.2 Chelate Method
2.4.3 Pechini Method
2.5 Spin Coating
3. EXPERIMENTAL METHODS
3.1 Synthesis of Nanoparticles
3.2 Synthesis of Thin Films
4. CHARACTERIZATION TECHNIQUES
4.1 X-ray Diffraction (XRD)
4.2 Electron Microscopy
4.2.1 Transmission Electron Microscopy (TEM)
4.2.2 Scanning Electron Microscopy (SEM)
vii
i
iii
v
vii
ix
1
1
2
3
3
4
5
8
9
11
11
11
12
15
15
16
17
17
18
19
19
20
21
21
22
25
25
27
27
29
4.3 Vibrational Spectroscopy (FT-IR, FT-Raman)
4.4 X-ray Photoelectron Spectroscopy (XPS)
4.5 Magnetic Measurements
4.6 Magnetic Resonance Imaging (MRI)
4.7 Quantum Chemical Calculations
5. RESULTS AND DISCUSSION
5.1 Nanoparticle Synthesis (Papers I-IV)
5.1.1 XRD
5.1.2 TEM
5.1.3 IR and Raman
5.1.4 XPS
5.1.5 MR
5.1.6 Magnetic Properties
5.1.7 Quantum Chemical Calculations
5.2 Thin Film Synthesis (Papers V and VI)
5.2.1 Na0.5K0.5NbO3 (NKN)
5.2.2 GdFeO3 & Gd3Fe5O12
6. CONCLUSIONS
ACKNOWLEDGMENTS
REFERENCES
PAPERS I-VI
viii
30
31
32
32
38
39
39
39
43
44
46
47
49
50
52
52
54
61
63
65
ABBREVIATIONS
Chemical compounds:
AcAc
Acetylacetone (C5H8O2)
AHDA
16-aminohexadecanoic acid (H2N−C15H30−COOH)
CA
Citric acid (C6H8O7)
DEG
Diethylene glycol (HOCH2CH2OCH2CH2OH)
EG
Ethylene glycol (HOCH2CH2OH)
FA
Formic acid (HCOOH)
GdIG
Gadolinium iron garnet (Gd3Fe5O12)
HHDA
16-hydroxyhexadecanoic acid (HO−C15H30−COOH)
NKN
Na0.5K0.5NbO3
OA
Oleic acid (CH3−C7H14−CH=CH−C7H14−COOH)
PEG
Polyethylene glycol (HO−(CH2CH2O)n−H)
TBAB
Tetrabutylammonium bromide ((C4H9)4NBr)
THF
Tetrahydrofuran (C4H8O)
TOPO
Trioctylphosphine oxide ((C8H17)3PO)
YAG
Yttrium aluminium garnet (Y3Al5O12)
YIG
Yttrium iron garnet (Y3Fe5O12)
Other:
ccp
EDX
FC
fcc
FT
FWHM
hcp
HREM
IR
IR
MRI
NCT
NMR
RF
RT
SAED
Cubic close-packing
Energy dispersive x-ray spectroscopy
Field-cooling
Face centred cubic
Fourier transform
Full width at half maximum
Hexagonal close-packing
High resolution electron microscopy
Infrared
Inversion recovery
Magnetic resonance imaging
Neutron capture therapy
Nuclear magnetic resonance
Radio frequency
Room temperature
Selected area electron diffraction
ix
SE
SEM
TE
TEM
TR
XPS
XRD
XRPD
ZFC
Spin-echo
Scanning electron microscopy
Echo time
Transmission electron microscopy
Repetition time
X-ray photoelectron spectroscopy
X-ray diffraction
X-ray powder diffraction
Zero-field-cooling
x
Introduction
1. INTRODUCTION
1.1 Materials Chemistry
In nature, rocks and minerals are mainly built from mixed metal oxides of
numerous crystal structures, where the atoms or ions are packed together in
a regular manner. Depending on their chemical composition and structure,
materials will possess different properties with respect to e.g. electrical and
thermal conductivity, optical and magnetic properties, and so on. Even
before classical antiquity, man had the knowledge of forming hard ceramic
materials from clay (pottery), of glass-making and also knew how to use
colour pigments derived from naturally occurring inorganic materials.
Though the understanding of the underlying chemistry was very poor, the
manufacturing and utilization of various materials were highly developed.
Today, aided by a battery of sophisticated analytical tools, such as high
resolution electron microscopy, x-ray diffraction and various spectroscopic
techniques, often used in combination with quantum chemical calculations,
we have the possibility to describe in detail how materials are built at an
atomic level, both with respect to chemical composition and structure. We
also know a great deal of how material properties are related to structure and
composition.
Based on this knowledge, novel materials with new and interesting
properties can be made by chemical or physical methods. One of the
challenges for materials chemistry today is to prepare these compounds, as
well as new ones, at the nanometre scale, as nanocrystals, nanofibres or thin
films, whereby novel properties of the materials can be achieved. Often
other synthesis techniques than the conventional solid state methods have to
be utilized, e.g. colloidal or sol-gel methods. The emergence of so called
nanochemistry has therefore brought about renewal of colloid chemistry,
resulting in the development of several important synthesis techniques [1].
These chemical methods are often referred to as bottom-up processes, implying
the structures to be built up from single atoms or ions, as compared to the
top-down processes, in which nanostructures are carved or etched from larger
bodies.
1
Introduction
1.2 Nanoparticles
Today, nanoparticles are used in a number of applications, ranging from
bio-medical products, to electronic devices and “everyday” products such as
ski wax, skin care products, stain repellent, wrinkle resistant threads etc.
Whether or not all these products can be regarded as “important” is not an
issue to be discussed here. Nanomaterials are usually defined as materials
having at least one dimension which is < 100 nm. For very small crystals, a
considerable fraction of the atoms are located at the surface of the crystal.
For example, a 5 nm spherical crystal has about 30 % of its atoms located at
the surface, as compared to bulk materials, where less than 1 % of the atoms
are found at the surface. The high surface to volume ratio is a fundamental
characteristic of nanomaterials, determining many of their properties.
Furthermore, due to the laws of quantum physics, electrical and magnetic
properties of materials are to some extent size-dependent, implying that for
very small crystals these properties may be different from those encountered
in the bulk [2]. Hence, nanocrystals might exhibit properties characteristic of
both the atomic and bulk states. The Royal Society [3] has proposed
Nanoscience to be defined as the study of
phenomena and manipulation of materials at atomic, molecular and
macromolecular scales, where properties differ significantly from those at
a larger scale
The possibility of obtaining new material properties at the nanoscale is the
main reason for the enormous interest in nanoscience and nanotechnology
in recent years. For a nanocrystal with a diameter of only a few nanometres,
the electronic structure of the crystal consists of more or less discrete energy
levels rather than of bands, implying that even pure metals such as gold
behave as semiconductors at this scale. Nanocrystalline magnetic materials
may exhibit superparamagnetism, a purely nanoscale phenomenon, where
the coupling of the atomic magnetic moments in a ferro- or ferrimagnetic
material vanishes in the absence of an external magnetic field. Such
superparamagnetic particles are today used as contrast enhancing agents in
magnetic resonance imaging (MRI), in the form of, e.g. magnetite (Fe3O4)
nanocrystals. Usually iron oxide contrast agents produce a negative contrast,
and it is regarded as an important task to develop nano-sized contrast agents
for positive contrast enhancement. Interesting compounds in this respect
2
Introduction
are gadolinium containing oxides, e.g. Gd2O3 and GdFeO3, both of which
will be discussed in this thesis.
If rare earth containing nanocrystals are to be used as contrast agents
in MRI on patients, a number of requirements have to be fulfilled. The most
important requirement is that the particles must not exceed about 5 nm in
size, and have a narrow size distribution. Further, the particles must be
biocompatible, i.e. accepted by the body, and water soluble. It is also
desirable that the particles can be made tissue-specific (e.g. for cancer cells)
by appropriate molecular coating. Obviously, these requirements make the
synthesis of such particles a challenging task.
1.3 Thin Films
Thin ceramic films ranging from tens of nanometres up to microns, have
many uses in today’s technology. Films of very hard materials, e.g. TiN, are
used as coatings on cutting tools, improving their resistance to heat and
abrasion, thereby extending the lifetime of the tool [4]. Thin films of
materials with specific electric and/or magnetic properties, are used in gas
sensors [5], high frequency transducers [6] or data storage devices [7].
Common methods to produce thin films today are sputtering techniques,
various chemical vapour deposition methods, and the sol-gel technique. The
latter synthesis technique has the advantages of being relatively cheap, can
be applied at rather low temperatures, and does not require expensive highvacuum equipment.
1.4 Crystal Structures
Many inorganic structures are named after minerals found in nature, e.g.
spinel, sphalerite, corundum, ilmenite, perovskite, garnet etc. Crystalline
solids are usually thermodynamically more stable than amorphous ones, with
high lattice energies making them more resistant to heat and dissolution.
Compounds consisting of two or more elements, more often than not,
occur in more than one crystal structure (polymorphism) and most elements
in the Periodic Table also exhibit more than one crystalline form in the solid
state (allotropes). One well-known example is carbon which occurs in many
different amorphous forms, as well as in the crystalline allotropes graphite
and diamond. The carbon allotropes illustrate the close relationship between
chemical composition, crystal structure and physical properties. Diamond
which forms a three-dimensional covalent network is the hardest substance
known and is electrically an insulator. Graphite, on the other hand, has a
3
Introduction
layered hexagonal structure, which is rather soft and has good electrical
conductivity. In nanocrystal or thin film synthesis, one usually strives to
obtain highly crystalline samples, in order to control the stability and
physical properties of the material.
1.4.1 Sesquioxides
The trivalent rare earths form the sesquioxide RE2O3 all of which are basic,
and insoluble in pure water, (the stability constants of the compounds are in
the order of 10–31). Principally three types of structures are adopted by the
sesquioxides. The early lanthanides form the A-type structure built from
octahedral-like REO7 units. The lanthanides in the middle form the
monoclinic B-type structure, also consisting of REO7 units but with another
geometry. The middle and late lanthanides form the C-type structure, which
is related to the fluorite structure but with the coordination of the metal
atom reduced from 8 to 6, as one quarter of the anions are removed in the
structure. Gd2O3 and Y2O3 are C-type oxides with the cubic structure (space
group Ia 3 (No. 206)) with a = 10.79 Å and 10.604 Å, respectively. The
structure is shown in Fig. 1.1.
Figure 1.1. The structure of C-type RE2O3 seen along [111] ∗ .
Red = O, green = RE containing polyhedra.
∗
The structure in this, and the following crystal structure images, have been generated
using the DIAMOND software version 3.1.
4
Introduction
All rare earths (except Yb) are paramagnetic, due to the presence of
unpaired f and/or d electrons. Nanocrystals of the oxides of these elements,
e.g. Gd2O3 or Dy2O3, might therefore be interesting as contrast agents in
magnetic resonance imaging (MRI). Y2O3 being isostructural with Gd2O3 is
quite similar in many ways, both having large negative standard redox
potential values for the half-reaction M3+ + 3e– → M(s) (−2.29 V and
−2.37 V for M = Gd and Y, respectively), as well as similar melting points
(2242 °C and 2439 °C, respectively). Y2O3 is diamagnetic, though, and in
this work it is used as a reference system in computational studies of Gd2O3.
The Gd3+ ion is strongly paramagnetic due to its seven unpaired 4f
electrons, corresponding to a (theoretical) magnetic moment of 7.94 BM.
Except for the noble gas nuclide 135Xe, gadolinium is the element with the
highest cross section value for thermal neutrons (2.55⋅105 and 6.10⋅104 barns
for 157Gd and 155Gd, respectively) making it a most interesting candidate in
neutron capture therapy (NCT) as well [8]. On the other hand, being
regarded as to some extent toxic, the in vivo use of gadolinium has to be
limited to low doses.
In a recent study by Bridot et al. [9], it was shown that ultra-small
Gd2O3 nanoparticles (∼2.2 nm) coated with polysiloxane and with an outer
layer of a low weight polyethylene glycol (PEG), were secreted into the
bladder about an hour after injection in a mouse. The study indicates that
there is a reasonably good prospect for gadolinium containing
nanocrystalline oxides to be used as contrast agents in in vivo MRI.
1.4.2 Perovskites
One technically very important mixed metal oxide structure is the perovskite,
named after the mineral CaTiO3. The mineral was discovered in 1839 by the
German mineralogist Gustav Rose (1798-1873) in the Ural Mountains. He
named the mineral in honour of his Russian colleague Count Lev
Aleksejevitj Perovskij (1792-1856) [10]. Perovskites have the general formula
ABO3 where A is a larger cation than B. In CaTiO3, which is orthorhombic,
the B/A radius ratio is ~0.65. The ideal cubic perovskite is built up by
corner sharing BO6 octahedra, with the A atom in 12-coordinated cavities in
between the octahedra. In Fig. 1.2 the structure of perovskite is represented,
both the cubic (left) and a slightly distorted one (right). The symmetries of
the distorted structure are naturally lower, e.g. orthorhombic. The perovskite
structure is the basis for high temperature superconductors, as well as a
whole range of dielectric ceramics.
5
Introduction
Figure 1.2. The structure of an idealized (left) and a distorted (right) perovskite.
Red = O, grey = A atom, turquoise = B containing octahedra.
An interesting ceramic material with a slightly distorted perovskite structure
is sodium potassium niobate, Na0.5K0.5NbO3 (NKN). It is usually described
as orthorhombic with a = 3.994 Å, b = 4.016 Å, c = 3.935 Å [11].
However, it has also been indexed as monoclinic with the β angle
close to 90° [12]. Pure synthetic NaNbO3 a is often described as
orthorhombic (space group Pbcm (No. 57) with a = 5.506 Å, b = 5.566 Å,
c = 15.520 Å) [13], while KNbO3 is usually described as tetragonal (P4mm
(No. 99) with a = 4.063 Å, c = 3.997 Å) [14]. In Table 1.1 some of the
suggested symmetries and unit cell parameters for NaNbO3, KNbO3 and
NKN are given.
NaNbO3 is antiferroelectric while KNbO3 is ferroelectric. The two
compounds form a continuous solid solution, Na1–xKxNbO3, which
becomes ferroelectric already for small incorporations of potassium [15].
The radius ratios r(NbV)/r(Na+) and r(NbV)/r(K+) are ~0.67 and ~0.51,
respectively. The NKN structure is non-centrosymmetric, and it possesses
piezoelectric properties. NKN has been suggested a candidate for use in, e.g.
non-volatile memories and electrically tuneable devices (see Papers VIII and
IX and references [12,16-18] for details). As non-toxic, NKN is also
considered for biomedical applications, where it constitutes an alternative to
PZT ceramics Pb(Zr,Ti)O3, which contains toxic lead.
a
The two minerals of NaNbO3, Lueshite and Natroniobite are described as, respectively,
orthorhombic (P2221 (No. 17), a = 5.51 Å, b = 5.53 Å and c = 15.50 Å), and monoclinic
(P2/m (No. 10), a = 3.9114 Å, b = 3.8813 Å, c = 3.9114 Å and β = 90.67°).
6
Introduction
Table 1.1. Some of the suggested symmetries and unit cell parameters
for NaNbO3, KNbO3 and NKN.
Compound
Symmetry
a (Å)
b (Å)
c (Å)
β(º)
Ref.
NaNbO3
Orthorhombic
3.950
3.875
3.914
-
[11]
Orthorhombic
5.506
5.566
15.520
-
[13]
Orthorhombic
5.598
15.523
5.505
-
[19]
Monoclinic
3.915
3.881
3.915
90.73
[20]
Orthorhombic*
5.572
3.881
5.501
-
-
Tetragonal
4.063
4.063
3.997
-
[14]
Orthorhombic
4.027
4.057
3.914
-
[11]
Monoclinic
4.038
3.971
4.038
90.25
[20]
Orthorhombic*
5.722
3.971
5.697
-
-
Orthorhombic
3.994
4.016
3.935
-
[11]
Monoclinic
3.99
3.93
3.99
90.35
[12]
Orthorhombic*
5.66
3.93
5.63
-
-
KNbO3
NKN
15.524#
15.884#
15.72#
* Shirane et al. [20] report that the relationships between the monoclinic (a’, b’, c’ and β)
and the orthorhombic (a, b, c) cell parameters are given by a = 2a’sin(β/2), b = b’,
c = 2a’cos(β/2).
# If b is multiplied by 4, the parameters agree better with those found by Lanfredi et al. [19].
The gadolinium orthoferrite GdFeO3 has a distorted perovskite structure, in
which the coordination number of the large A cation (Gd3+) is reduced from
12 to 8. This can be explained by the less favourable ionic radius quotient
r(Fe3+)/r(Gd3+) ≈ 0.72. GdFeO3 is orthorhombic (Pbnm (No. 62)) with
a = 5.349 Å, b = 5.611 Å and c = 7.669 Å.
GdFeO3 has interesting opto-magnetic properties, and has been
studied for several decades [21-27]. The over-all magnetic behaviour of bulk
GdFeO3 is described as the result of two contributing magnetic “sublattices”:
(i) an antiferromagnetic iron oxide lattice in which the spins are coupled via
an Fe3+−O2−−Fe3+ superexchange mechanism; and (ii) a paramagnetic
contribution from essentially non-coupled Gd3+ ions. Due to spin-canting in
7
Introduction
the iron containing sublattice, as a result of the distorted perovskite
structure, a small ferromagnetic moment is observed in one particular
crystallographic direction [22,23]. GdFeO3 has been suggested for several
different applications, e.g. in gas sensors, or data storage devices such as
bubble-domain memories. In nanocrystalline form, it might be a candidate
for use as a contrast agent in MRI, as discussed in Paper IV.
1.4.3 Garnets
The garnet structure, with the general formula A3B2X3O12, is adopted by
several binary or higher oxides. In this structure A is usually a fairly large
eight-coordinated cation, while B and X are somewhat smaller cations with
respectively octahedral and tetrahedral coordination. The garnet structure is
shown in Fig. 1.3. One important garnet phase is the yttrium iron
compound Y3Fe5O12 (YIG), which has a body centred cubic cell with
a = 12.38 Å ( Ia3d (No. 230)) [28]. In the YIG structure, yttrium is
positioned in the A sites, and iron(III) are found in both the B and X sites,
forming two iron−containing sublattices. YIG is an interesting material for
microwave communication devices (cell phones and satellites) [29]. Another
frequently used garnet is the yttrium aluminium garnet (YAG), where the
iron is replaced by aluminium. If about 1 % Nd3+ is substituted into the
structure, it can be used as laser material for the emission of infrared
(1064 nm), or green (532 nm) light.
If in YIG, yttrium is substituted for gadolinium, the garnet Gd3Fe5O12
(GdIG) is obtained. The cell parameter for GdIG is slightly larger than for
YIG (12.47 Å), but the symmetry is the same [30,31]. GdIG is ferrimagnetic
with a magnetic moment of ~16 BM per formula unit. In similarity with
other rare earth iron garnets, the magnetic moment shows an unusual
temperature dependence. With increasing temperature, the magnetic
moments vanish at the so called compensation temperature, Tcomp. When the
temperature is further raised, the magnetic moment increases in the opposite
direction and again turns towards zero at the Curie temperature. The effect
is believed to arise from the existence of two Fe3+ sublattices, where the
spins of the sublattices do not randomize at the same speed [32].
B
8
Introduction
Figure 1.3. The garnet structure, seen along [100].
Red = O, grey = A atom, turquoise = B, X containing polyhedra.
1.4.4 Phase Diagram of the System Fe2O3-Gd2O3
According to the 1950’s investigation by Warshaw and Roy, there are only
two stable phases in the system Fe2O3-Gd2O3, namely the perovskite
GdFeO3 and garnet Gd3Fe5O12 phases [33]. In their work, they used
hydroxide precipitates with different metal stoichiometries, which were
subsequently calcined at elevated temperatures. This rather barren chemical
landscape was unexpected, and a number of compounds one would have
thought possible such as GdFe2O4, Gd2Fe3O7 etc. did not show up and were
assumed to be metastable (if existing at all). In Fig. 1.4 the phase diagram of
the Fe2O3-Gd2O3 system is shown for temperatures 800-1200 °C. Notably,
the Gd2O3 transforms from the cubic C-type to the monoclinic B-type
structure at ~1030 °C, in the Gd2O3 rich part of the diagram.
However, Thakur et al. [34] reported an orthorhombic phase with the
composition GdFe2O4 (marked with  in Fig 1.4) prepared by solid state
synthesis, though no space group of it was given. The cell parameter were
found to be a = 6.242 Å, b = 7.366 Å and c = 8.836 Å. Among the heavier
lanthanides, e.g. Er, Yb and Lu, as well as for Y, a two-dimensional
antiferromagnetic phase adopting the AB2O4 composition is known. The
9
Introduction
phase has a trigonal symmetry with a short a axis and a long c axis, (e.g. for
YFe2O4 a = 3.5136 Å, c = 24.7781 Å, s.g. R3mH No. 166) [35].
1200
1000
Fe2O3
Gd3Fe5O12
+ Fe2O3
20
GdFeO3 + B-Gd2O3
Gd3Fe5O12
+ GdFeO3
Temp. (°C)
GdFe2O4
40
GdFeO3 + C-Gd2O3
60
Mole % Gd2O3
80
Gd2O3
Figure 1.4. Part of the Fe2O3-Gd2O3 phase diagram, after Warshaw and Roy [33].
In the Ga2O3-Gd2O3 system, which is similar to that of Fe2O3-Gd2O3 (Ga3+
and Fe3+ are of about the same size), additional phases have been found.
Nicolas et al. [36] found, besides the garnet phase, compounds with the
compositions Gd3GaO6 and Gd4Ga2O9, while the perovskite compound
GdGaO3, being more difficult to synthesize, was reported by Guitel et al.
[37]. Using physical preparative methods, there have been several reports of
a hexagonal form of REFeO3 for most rare earths, as thin films or
nanocrystals [38-40].
According to the rule by Ostwald, the thermodynamically least stable
state is the one formed first when a solid crystallizes [41,42]. It is not
unlikely, therefore, that such (metastable) phases may appear in nanocrystal
and thin film synthesis, as the formation of such materials to a large extent is
governed by kinetics instead of thermodynamics.
10
Introduction
1.5 Material Properties
1.5.1 Electrical Polarization
In a parallel plate capacitor with vacuum between the metal plates, the
capacitance C0 is defined by:
C0 = ε0A/d
(1.1)
where ε0 is the vacuum permittivity (8.854⋅10–12 F⋅m–1), A is the area of the
plates (m2), and d is the distance between the plates (m).
An insulating material with zero net dipole moment can be polarized if it is
placed in an electric field. An opposed field is induced in the solid arising
from the distortion of the electron clouds in the electric field and/or the
slight displacement of the atoms/ions from their equilibrium positions in
the lattice. The induced dipole moment is proportional to the applied field.
Usually the polarization is described by the relative permittivity, εr, which
can be determined by measuring the capacitance of an electric circuit in the
presence and absence of the (dielectric) solid, i.e.
C/C0 = εr
(1.2)
where C and C0 are the measured capacitances (in farad) with and without
the solid, respectively.
1.5.2 Piezo- and Ferroelectricity
An insulating or semiconducting crystal belonging to a non-centrosymmetric
point group often possesses piezoelectricity. If an electric field is applied
across a piezoelectric crystal, a mechanical strain develops more or less
normal to the electric field, and vice versa, if a pressure is applied across the
piezoelectric crystal an electrical polarization of the material is induced. The
piezoelectric phenomenon depends on both the crystal structure and the
direction of the applied stress.
A sub-group to piezoelectric materials are those possessing
ferroelectric properties. A ferroelectric material inherently possesses
electrical polarized domains, which in an applied electric field may be
reoriented and aligned. After removing the field, the ferroelectric material
remains polarized, (in contrast to paraelectric materials where the
polarization disappears when the field is zero).
11
Introduction
1.5.3 Magnetic Properties
If a paramagnetic material is placed in a magnetic field, the magnetic
moment associated with the spins of unpaired electrons will align
themselves more or less with the field, implying that the density of the field
lines will be higher inside the material than outside. When the field is
removed, the spins will resume a random orientation. The paramagnetic
effect is dependent on both the strength of the applied magnetic field, the
temperature, and on the material itself. For a truly paramagnetic material,
Curie’s law is obeyed, i.e.
χ=
C
T
(1.3)
where χ is the magnetic susceptibility, T is the absolute temperature, and C is
the Curie constant. However, in some materials the spins of nearby atoms
are “coupled”, either in parallel (ferromagnetism), or anti-parallel
(antiferromagnetism). The coupling will sustain up to a certain temperature
when a transformation to the paramagnetic state occurs. For ferromagnetic
materials, the transition temperature is called the Curie temperature, TC, and
for antiferromagnetic, the Néel temperature, TN. For such materials in their
paramagnetic state, a somewhat modified form of Curie’s law often is
obeyed (the Curie-Weiss law)
χ=
C
T +θ
(1.4)
where θ, the Weiss constant, is < 0 for antiferromagnetic materials and > 0
for ferromagnetic ones. In some antiferromagnetic materials the magnetic
spins do not cancel, i.e. a permanent net magnetic moment still remains in
the material (ferrimagnetism). Such materials are technically important as, e.g.
magnetic memory devices. As was mentioned above, for nano-scale
magnetic materials a phenomenon called superparamagnetism is sometimes
observed. Superparamagnetism occurs when in a permanent magnetic
material the crystallite size becomes smaller than that of a magnetic single
domain. In such (nanocrystalline) materials, the magnetic moment of the
crystal vanishes in the absence of an external magnetic field. Since
superparamagnetic particles do not aggregate as a result of interparticulate
magnetic forces when there is no external field, they have become
interesting for several biomedical applications, e.g. for magnetic particle
assisted drug-delivery to tumours, contrast enhancing agents in MRI etc.
12
Introduction
Bulk Gd2O3 is, basically, antiferromagnetic with rather low TN
(< 11 K; see Fig. 1.5). The magnetic properties of nanocrystalline gadolinia
are described in Paper III.
1800000
-1
-3
χ (mol · m )
1200000
600000
0
0
100
200
300
T (K)
−1
Figure 1.5. The inverse molar magnetic susceptibility, χ M , vs
temperature, T, for Gd2O3 powder in the temperature range
11-300 K. The measurement was performed on a weak-field acsusceptometer (Lake Shore 7100; 125 Hz, 250 A⋅m–1). The value of
the Curie constant obtained was C = 1.919⋅10–4 K⋅m3⋅mol–1,
corresponding to μeff = 7.81 Bohr magnetons per Gd3+, slightly
lower than the theoretical value (7.94 BM). Within the measured
temperature interval the behaviour of Gd2O3 is paramagnetic and
the overall behaviour antiferromagnetic with θ = −14.8 K [43].
13
Introduction
14
Synthesis Methods
2. SYNTHESIS METHODS
2.1 Colloidal Synthesis Methods and Nanochemistry
The word colloid is derived from the Greek word for glue, κόλλα, and
colloidal systems have been studied for more than a century. A colloidal
solution differs from a “true” one in that the dissolved components are not
single ions or molecules, but small particles with diameters ranging from
1 nm to ~500 nm. In 1857, Michael Faraday (1791-1867) prepared colloidal
gold and showed that light, when passed through such a solution is scattered,
a phenomenon now known as the Faraday-Tyndall effect. Faraday also
observed that solutions of colloidal gold can assume different colours, and
proposed that the phenomenon was due to the different sizes of the
particles [44]. Remarkably, at least one of Faradays original colloidal
preparations survived until World War II, when it was accidently lost by
“enemy action” [45].
In connection with nano synthesis, the basic idea of the colloidal
method is to limit particle growth. As the formation of very small crystals is
thermodynamically unfavourable due to the high surface energy of such
particles, nanosynthesis must be controlled by kinetics. To achieve this,
surfactant molecules are used to limit particle growth. One method uses a
non-polar solvent, leading to the formation of reversed micelles. In the polar
core of the micelle, a minute amount of water is present in which
precipitation of the nanocrystal takes place. The size of the micelles thus
determines the size of the particles. In Fig. 2.1 an illustration of nanoparticle
formation in reversed micelles is shown.
H2O
Mn+
NP
Oil
Oil
Figure 2.1. Reversed micelles in nanocrystal synthesis.
15
Synthesis Methods
Surfactant molecules may also act as capping molecules, preventing particle
growth. Without the presence of surfactants, the particles tend to grow, and,
as pointed out by Wilhelm Ostwald (1853-1932) a long time ago, larger
particles will grow at the expense of smaller ones, an effect known as
Ostwald ripening [46]. The colloidal method has become very popular and is
frequently used for the synthesis of pure metal nanoparticles,
semiconducting binary particles such as CdSe, and many other materials
with controlled size, shape, composition and structure [1,47-54].
It has been demonstrated that colloidal chemistry can be used to
build superlattices by self-assembled nanocrystals. The superlattice consists
of particles of the same kind and size in a close-packed manner [55-58], or
by particles of different sizes or composition yielding a diversity of
structures [59]. The most common packing of equally sized spheres is the
cubic close packing (ccp) with an fcc lattice, but in the case of nanocrystals the
less stable hcp is often obtained. The latter phenomenon has been explained
as a result of dipole-dipole interactions between the particles [60].
In the preparation of sub-10 nm nanoparticles of europium oxide
(Eu2O3) and terbium oxide (Tb2O3) Wakefield et al. [61,62] used
trioctylphosphine oxide (TOPO) as a capping agent. Another very
important method to prepare rare earth oxide nanoparticles was introduced
by Bazzi et al. [63], based on a method previously developed by Fievet et al.
[64] to produce metallic powders of transition metals. The process uses a
polyol, e.g. diethylene glycol (DEG), a high boiling point solvent (b.p. 244 °C
for DEG) which also act as capping molecule for the nanoparticles formed.
The method can be used to prepare a number of metal oxide compounds,
and was used in this thesis in the preparation of RE2O3 (RE = Gd, Y) and
GdFeO3 (Papers I-IV) [65]. Sometimes long-chained carboxylic acids, e.g.
oleic acid, are added to further stabilize the particles and make them soluble
in non-polar solvents [66]. The substitution of one type of capping molecule
for another one in order to change the solubility and/or physical or
chemical properties of the particles was demonstrated by Boal et al. [67] and
is here discussed in Paper I.
2.2 Electrochemical Methods
In recent years electrochemical methods have become popular in
nanoparticle synthesis [68-70]. In these methods a sacrificial anode is used as
metal source, which upon application of a positive potential, oxidizes to
cations. At the cathode the cations are again reduced, apparently forming
nano-sized metal clusters. As electrolyte a quaternary ammonium salt, e.g.
16
Synthesis Methods
tetrabutylammonium bromide (TBAB) dissolved in an organic solvent, such
as 2-propanol, THF, acetonitrile etc., is commonly used. The ammonium salt
serves both as charge carrier in the electrolyte and as capping molecule,
implying that the positively charged ammonium ions adsorb on the surface
of metallic nano-clusters formed facilitating their detachment from the
cathode. Both pure metal nanoparticles, e.g. Au, Pd, Ag, and metal oxides
have been synthesized by this technique. In the latter case, the oxidation can
be performed by bubbling air through the solution. The method has been
used for the preparation of small (~5 nm) ZnO nanoparticles with narrow
size distribution, as described in Paper XII (not included in this thesis).
2.3 Combustion Methods
A somewhat different method to produce nanoparticles is the combustion
method. In this method metal nitrates are mixed together with glycine
(H2NCH2COOH) in water. The water is evaporated by boiling and when
sufficiently dry the remaining slurry self-ignites, producing a very fine metal
oxide powder. The reaction is thus very rapid, and it is believed that the
short reaction time limits the crystal growth. The particle size can be
controlled by the glycine to nitrate ratio yielding crystals in the range
5-200 nm [71].
The reaction taking place is suggested to be
6 M(NO3)3 + 10 H2NCH2COOH + 18 O2 →
3 M2O3 + 20 CO2 + 5 N2 + 25 H2O + 18 NO2
The method is very suitable for preparing nanocrystalline metal oxide
powders. However, as there is no capping molecule involved in the synthesis,
the nanoparticles formed tend to aggregate, making it hard to disperse the
powder homogeneously in solution.
2.4 Sol-Gel Methods
Several more or less different chemical methods are incorporated under the
name sol-gel, but the basic objective of all of them is to achieve
homogeneous systems with respect to the desired metals. The sol-gel
methods, or “chimie douce”, was developed in the 1950s and 60s [72-74], but
the first reported studies of sol-gel synthesis goes back to the middle of the
nineteenth century [75-77]. By definition, a sol is a colloidal dispersion of
17
Synthesis Methods
small solid particles in a liquid, and a gel is a pseudo-solid where the solvent
is dispersed in a polymeric network. A simple description of the sol-gel
synthesis route is that a sol is prepared by dissolving metal ion complexes in
a suitable organic solvent, and, upon hydrolysis, the sol forms as a colloidal
metal oxide/hydroxide precipitate. After ageing and/or heating, the sol is
transformed into a gel in which the metal oxide or hydroxide particles form
a polymeric network enclosing the solvent. When the gel is heated at higher
temperatures, the organics evaporate or decompose, allowing the inorganic
solid to crystallize, (Fig. 2.2).
Heating
Sol
or ageing
Heating at high
Gel
temperatures
Solid
Figure 2.2. The basic steps of the sol-gel method.
The sol-gel technique encompasses a variety of methods. Generally speaking,
the choice of method depends on the desired properties of the material to
be prepared. Some methods are suitable both for making nanocrystals and
thin films. A few of these are described below.
Thin film synthesis with the sol-gel method includes at least four
steps: (i) preparation of a precursor solution; (ii) deposition of the solution
onto a substrate, (e.g. by spin coating); (iii) drying and subsequent pyrolysis
of the organics (~350 ºC), and (iv) heat treatment at higher temperatures for
crystallization of the film (500-900 ºC).
2.4.1 Alkoxide Method
A widely used sol-gel method is the one based on metal alkoxides. The
reagents in this method are metal alkoxides, M(OR)x, dissolved in an alcohol.
Often ethanol or some other common alcohol is used but occasionally
alcohols such as 2-methoxy ethanol (CH3OCH2CH2OH) are used. The
alkoxide solution is partially hydrolysed by adding water to it, according to
the reaction formula
M(OR)x + yH2O → M(OR)x–y(OH)y + yHOR
The hydroxy-alkoxide eventually forms a polymer containing an oxygenmetal-oxygen bonded network. Upon calcination of the gel, the metal oxide
18
Synthesis Methods
forms. As the alkoxide precursors are strongly hygroscopic, the synthesis
often has to be carried out in closed vessels.
2.4.2 Chelate Method
In the chelate method, which resembles that of the alkoxide method,
chelating ligands are used to stabilize the metal ions. Instead of the rather
expensive metal alkoxides the cheaper chlorides or nitrates are commonly
used, allowing the synthesis to be performed in open vessels. As chelating
agents, carboxylic acids, acetylacetone, amines etc. (see Table 2.1) are
frequently used.
Table 2.1. Some of the chelates used in this work.
Name
Structure
O
Acetylacetone
Name
Citric acid
O
Structure
COOH
HO
COOH
COOH
O
COOH
Oxalic acid
Acetonylacetone
COOH
O
O
O
Rochelle salt
O
3-n-amyl-2,4pentanedione
HO
HO
Na
O
O
O K
Succinic acid
+
+
COOH
COOH
2.4.3 Pechini Method
Another method, developed and patented by Pechini [78], is based on citric
acid and ethylene glycol. Usually the solvent in this method is water, and
metal chlorides or nitrates are used as the metal sources. The function of
citric acid is to chelate the metal cations. When ethylene glycol is added, it
reacts under heating with the citric acid in a polyesterification reaction,
forming metal-containing polymers. A flow chart of the synthesis procedure
is shown in Fig. 2.3. By varying the citric acid/ethylene glycol ratio, the
19
Synthesis Methods
viscosity of the solution can be controlled; an important factor in e.g. spin
coating.
Metal salts
in H2O
Citric acid
Heating
Polymerization
Deposition
(spin coating)
Ethylene
glycol
Heat treatment
(crystallization)
Figure 2.3. Steps in thin film preparation with the Pechini method.
2.5 Spin Coating
Spin coating is a common method for depositing solutions on flat substrates.
Other methods e.g. dip coating or spray coating, are often used for more
irregularly shaped substrates. In spin coating, after depositing a few drops of
the solution, the substrate is rotated at high speed, (ω = 1000-5000 rpm),
leaving a very thin layer of metal-containing solution on the surface of the
substrate, as illustrated in Fig. 2.4.
Solution
ω
Substrate
Deposition
Spinning
Thin film
Figure 2.4. The principle of the spin coating technique.
Mak et al. [79] have shown by ellipsometry that the concentration of the
deposition solution is important for quality and layer thickness of the sol-gel
derived film. Other parameters affecting the film thickness are the viscosity
of the solution, the liquid’s wetting of the substrate, and the spin speed and
time.
20
Experimental Methods
3. EXPERIMENTAL METHODS
3.1 Synthesis of Nanoparticles
Nanoparticles of Gd2O3, Dy2O3, Y2O3 and GdFeO3 were synthesized by
colloidal methods, and by the combustion method, both of which are
described in Chapter 2. In this work, the main technique used, was the
polyol method, and a typical synthesis proceeded as follows
In a certain amount of DEG, the metal chlorides or nitrates of the desired
metals are dissolved and heated to 100-140 °C. To this solution, NaOH or
KOH dissolved in DEG is added, further heated to 180-210 °C and
refluxed for about 4 hours.
Particles made by the combustion method were synthesized according to the
following procedure
Equal volumes of 100 mM solutions of metal nitrates and glycine are
mixed in an E-flask ending up with a nitrate/glycine ratio of 3:1. Under
heating and stirring, the water is evaporated resulting in a brownish slurry,
which upon further heating self-ignites. Highly crystalline nanoparticles are
thereby obtained.
See Papers I-IV for further details.
In addition to the nanocrystal syntheses, some solid state syntheses were
performed to obtain reference materials of mainly the mixed gadolinium
iron oxide compounds. In Table 3.1, the solid state syntheses performed are
summarized.
21
Experimental Methods
Table 3.1. Summary of the solid state syntheses.
No.
Intended compound
Starting materials
Attained compound(s)*
1
GdFeO3
Gd2O3, Fe2O3
GdFeO3
2
Gd3FeO6
Gd2O3, Fe2O3
B-Gd2O3, GdFeO3
3
Gd3GaO6
Gd2O3, Ga2O3
Gd3GaO6
4
Gd3Fe0.5Ga0.5O6
Gd2O3, Fe2O3, Ga2O3
Gd3Fe0.5Ga0.5O6
5
Gd3Fe0.75Ga0.25O6
Gd2O3, Fe2O3, Ga2O3
B-Gd2O3, Gd(Fe,Ga)O3,
Gd3(Fe,Ga)O6
* Determined by XRD. Note: Gd2O3 as starting material is cubic, while B-Gd2O3 stands for
monoclinic gadolinium oxide. The samples were heated at 1200-1400 °C for about 24 h.
3.2 Synthesis of Thin Films
In the synthesis of the Na0.5K0.5NbO3 (NKN) films, several different sol-gel
routes were attempted. The principle idea is to dissolve the metal-containing
salts in a proper solvent so that a sol is formed, as described in Chapter 2. A
typical modified Pechini route synthesis is performed as follows
Niobium pentachloride, NbCl5, (0.01 mol) is dissolved in a mixture of
acetylacetone, AcAc, (10 ml) and absolute ethanol (40 ml), and the solution
is heated under N2 purge at 60 °C for 30 minutes. Citric acid (0.1 mol) is
added to the hot solution, and pH is adjusted to ~4 with NH3 (aq).
Thereafter, potassium acetate, KOAc (0.005 mol) and sodium acetate,
NaOAc (0.005 mol) dissolved in acetic acid, are added, followed by
ethylene glycol (0.25 mol). The solution is stirred and heated at 60 °C for a
few hours to yield a final Nb concentration of ~0.17 M. The molar
Nb/Na/K ratio is 2:1:1, the niobium/citric acid ratio 1:10, and the citric
acid/ethylene glycol mass ratio 3:2.
The sols were spin coated onto substrates of either pure silicon wafers,
mixed platinum/iridium plates (Pt80Ir20), or layered wafers consisting of
platinum, titanium, silicon dioxide and silicon (Pt/Ti/SiO2/Si).
22
Experimental Methods
The GdFeO3 and Gd3Fe5O12 thin films were prepared by chelating the metal
ions by acetylacetone (AcAc) in 2-methoxy ethanol. A similar method was
previously used in the synthesis of α-Fe2O3 thin films [80]
Gadolinium and iron nitrates are dissolved in a 1:8 by volume mixture of
AcAc and 2-methoxy ethanol and with a metal concentration of 0.3 M. The
mixture is stirred for 2 hours at room temperature. The sol obtained is spin
coated onto silicon wafers, using a speed of 3000 rpm for 10 s. A total of
20 layers are deposited, and after each deposition the wafer is heated on a
hot plate to evaporate the solvents. After the final deposition, the films are
annealed in a tube furnace at 700 °C for 20 hours.
Details on the synthesis procedures are given in Papers V and VI.
23
Experimental Methods
24
Characterization Techniques
4. CHARACTERIZATION TECHNIQUES
4.1 X-ray Diffraction (XRD)
Intensity (arb. units)
Any solid material with a periodically ordered atomic or molecular structure
diffracts electromagnetic radiation of wavelengths of about the same length
as the interatomic distances. In x-ray diffraction, the wavelength is ~1 Å
(1.5406 Å for CuKα1 radiation). The diffracted radiation gives rise to specific
patterns (diffraction patterns), as shown in Fig. 4.1 for a powder sample of
GdFeO3. In principle, the diffraction patterns contain information of both
the size and symmetry of the unit cell, as well as the location of the atoms
within the cell. However, the latter information is hidden and has to be
mathematically extracted from the intensity of the reflections before an
electron density map can be constructed. A crystal structure, in the classical
sense of the word, can be described by one of 230 space groups, which
provide information about the crystal symmetries and atomic positions.
15
20
25
30
35
40
45
50
55
60
2θ (deg.)
Figure 4.1. The XRD powder pattern for GdFeO3.
25
65
Characterization Techniques
The relationship between the x-ray wavelength, λ, the interplanar distance of
the lattice planes hkl, dhkl, and the angle of the incident beam to the same set
of lattice planes, θ, is given by Bragg’s law, i.e.
2dhklsinθ= nλ
(4.1)
where n is the diffraction order (an integer > 0). As shown in the Footnote
below, Bragg’s law can be easily derived from simple geometrical
considerations a .
For very small crystallites such as nanocrystals, considerable peak
broadening occurs in powder diffraction. The reason for this is that for very
small crystallites, the number of parallel crystal planes hkl is too limited and
the Bragg condition therefore is not fulfilled. The peak broadening, however,
can be used to estimate the average size of the crystallites from the Scherrer
equation
t=
0.9 λ
(4.2)
B M2 − BS2 cosθ
In Eq. 4.2, t is the estimated crystallite size in Ångströms, and BM and BS are
the FWHMs (full width at half maximum) in radians for, respectively, a
strong peak in the sample, and a reference peak in a large-grained standard.
The standard is needed to correct for instrumental broadening. Additional
difficulties in measuring the exact crystallite size are due to crystal
imperfections, lattice strain etc., all of which contribute to the broadening of
the diffraction peaks.
B
B
a
Diffracted
beam
Incident
beam
θ
θ
θ
dhkl
s
Crystal
planes
s
Only when the diffracted waves fulfil the condition 2s = nλ, constructive interference
takes place. The amplitude, or structure factor Fhkl, of the diffracted beam is given by,
F = ∑ f exp(2 πi ( hx j + ky j + lz j )) where fj and (xj,yj,zj) are, respectively, the atomic
hkl
j
j
scattering factor and coordinates of the j th atom in the unit cell. The intensity of the
diffracted peak is proportional to the squared amplitude of the diffracted wave, i.e.
I hkl ∝ Fhkl
2
26
Characterization Techniques
The nanocrystals and thin films of this study were characterized by XRPD
on a Philips PW 1820 diffractometer using CuKα1 radiation (λ = 1.54056 Å).
The diffractograms were recorded with θ/2θ configuration, with a current
setting of 40 mA, a generator voltage of 40 kV, a divergence slit of ½º, and a
receiving slit of 0.2º.
4.2 Electron Microscopy
In electron microscopy, high energy electrons are generated from a filament
by using a large accelerating voltage (up to ~400 kV). According to the de
Broglie equation, the high momentum of such electrons corresponds to very
short wavelengths (λ is ~0.025 Å for an accelerating voltage of 200 kV).
Relativistic effects have to be considered when the accelerating voltage is
higher than ~100 kV, so the wavelength is calculated as
λ=
h
(4.3)
1
⎛
e 2U 2 ⎞ 2
⎜⎜ 2em 0U + 2 ⎟⎟
c ⎠
⎝
In Eq. 4.3, h is the Planck constant, c is the speed of light, e is the electron
charge, m0 is the electron rest mass, and U is the accelerating voltage.
4.2.1 Transmission Electron Microscopy (TEM)
Modern transmission electron microscopes can be used for a number of
sophisticated analysis techniques, but the two basic ones used most
frequently are object imaging and electron diffraction. Fig. 4.2 shows the
optical principles behind the two configurations from sample to screen.
When the electron beam is transmitted through the sample, the elastically
scattered electrons are passed through apertures and electromagnetic lenses
before projected onto the screen. When the selected area electron
diffraction (SAED) mode is in operation, a selected area aperture is inserted
into the beam, while in imaging mode an objective aperture is sometimes
used to enhance the contrast.
27
Characterization Techniques
Sample
Objective lens
Objective
aperture
Selected area
aperture
Intermediate lens
Projector lens
Diffraction
pattern
Image
Figure 4.2. The optical configurations used in TEM,
for electron diffraction (left), and sample imaging (right).
In electron diffraction, the scattering angle θ is very small (typically < 2 º)
and the d-values can be calculated from
d hkl =
Lλ
rhkl
(4.4)
where L is the camera length in mm, λ is the electron wavelength, and rhkl is
the distance between the central electron beam (000) and the diffraction
spot hkl in mm. The equation can be understood from the schematic
drawing in Fig. 4.3.
28
Characterization Techniques
Sample
1/λ
L
Ewald sphere
Reciprocal
lattice plane
1/dhkl
000
hkl
Film
rhkl
Figure 4.3. Since the Ewald sphere is almost “flat” in electron diffraction due
to the very short wavelength of the electron, it holds from simple geometry
that
1/λ
1/d hkl
=
L
r hkl
immediately yielding the above relationship in Eq. 4.4.
High resolution electron microscopy (HREM) is a very powerful tool for
structural studies. Nowadays a resolution better than 2 Å can be achieved, at
least for inorganic crystals. One of the advantages with the technique is that
the phase information of the structure factors is preserved in the image.
HREM is also a useful tool in the study of crystal defects.
In this study the sizes and crystallinity of the nanoparticle samples
were examined with a FEI Tecnai G2 microscope, operated at 200 kV.
4.2.2 Scanning Electron Microscopy (SEM)
The principal differences between TEM and SEM are that in the latter
technique the electron beam is scanned over the sample surface, instead of
being passed through the sample. The accelerating voltage used is also
considerably lower (usually 5-20 kV). When the electron beam hits the
atoms in the sample, both electrons and photons are emitted. The emitted
electrons are collected and used to form a 3D picture of the sample,
providing both topographical and structural information. Different kinds of
29
Characterization Techniques
electromagnetic radiation are emitted from the sample, and one of them is
x-rays. Since the emitted x-ray photons have energies characteristic for each
element, they can be used to determine the chemical composition of the
sample, qualitatively and quantitatively, by a technique called energy
dispersive x-ray spectroscopy (EDX).
The microstructure of the thin films was examined by scanning
electron microscopy (SEM), on a LEO Gemini 1550 FEG, equipped with
an Oxford LINK ISIS system with a Ge detector for EDX analysis.
4.3 Vibrational Spectroscopy (FT-IR, FT-Raman)
Vibrational spectroscopy was used to study the molecular coating of the
nanoparticles. For a non-linear molecule containing N atoms, there are
3N − 6 normal modes of vibration, and for a linear one 3N − 5 modes. For
crystal lattices, the number of vibrations is somewhat more complicated to
interpret. If the primitive lattice consists of σ molecules, which in turn
consist of N atoms, there will be 3 modes that are acoustical and 3Nσ − 3
that are optical. The optical mode is further divided into (3N − 6)σ internal
modes and 6σ − 3 lattice modes. The most common kinds of vibrational
modes are stretching, bending, twisting, wagging and rocking. Depending on
how stretching modes affect the symmetry of the molecule, it is classified as
antisymmetric or symmetric, while a bend may be degenerate or symmetric.
Only those normal vibrational modes accompanied by a change of the
dipole moment of the molecule are IR active.
If a carboxylic acid is chemisorbed onto a surface, three different
types of coordination modes are possible, namely the monodentate,
bidentate (chelating), or bridging modes, see Fig. 4.4. The wavenumber
separation, Δ, between the antisymmetric, νas(COO–) and the symmetric,
νs(COO–) stretching bands can be used to infer the coordination mode. If Δ
is large (> 200 cm–1) the carboxylate group usually acts as a monodentate
ligand, and if Δ is small (< 140 cm–1) the coordination is bidentate. However,
if the bidentate coordination is asymmetric, that is if the two bond lengths
are distinctly different, the Δ value tends to be higher, even approaching the
monodentate value. For a bridging ligand, the Δ value usually lies between
140 and 200 cm–1 [81,82]. The values of Δ are, however, not to be taken as
absolute indicators of the coordination mode but more approximate.
30
Characterization Techniques
R
R
R
O
O
M
O
O
M
O
O
M
M
Figure 4.4. Monodentate, bidentate and bridging coordination
of a carboxylic group to a metal.
IR and Raman spectroscopy are similar in so far as both techniques are
sensitive to molecular vibrations. While IR spectroscopy measures infrared
photons absorbed by molecular vibrations, Raman spectroscopy measures
infrared photons scattered by such vibrations. The monochromatic photons
used in a Raman experiment are generated by a laser. As already noted, an
IR photon will be absorbed only if its energy corresponds to a vibration,
where the dipole moment of the molecule is changed. A Raman photon, on
the other hand, will be scattered only if it gives rise to a change of the
polarizability of the molecule. For a centrosymmetric molecule, it holds that
every vibration which is observable in IR is invisible in Raman, and vice
versa (the rule of mutual exclusion). IR and Raman are thus in a sense
complementary techniques for the study of molecular vibrations.
In the present work FT-IR measurements in transmission mode were
performed on a Perkin-Elmer Spectrum 1000 spectrometer, using pressed
KBr pellets. Spectra of nanoparticles were obtained over the range
4000-400 cm–1, with a spectral resolution of 2 cm–1 using 16 scans. FT-IR
measurements in reflecting mode were performed on a BioRad QS-2200
FT-IR spectrophotometer at room-temperature, in the frequency range
800-400 cm–1, using 500 scans. Raman measurements were performed on a
Bruker IFS66 FT-Raman spectrometer, and the samples were excited with
an Nd-YAG laser (1064 nm).
4.4 X-ray Photoelectron Spectroscopy (XPS)
XPS is a surface sensitive technique of chemical analysis, based on the
photoelectric effect. The sample, spin-coated onto a flat substrate, is
exposed to monochromatic x-rays, causing photoemission from the valence
and core levels of atoms in the outermost atomic layers of the sample. The
kinetic energy, EK, of emitted photoelectrons from the valence band can be
determined by the Einstein equation
31
Characterization Techniques
EK = hν − EB + Φ
(4.5)
B
where hν is the x-ray photon energy, EB is the binding energy of the core
level, and Φ is the work function. Photoemission of valence electrons
provides information of the bonding in the solid, and emission of core
electrons of its chemical composition. Other parameters often determined
by XPS are oxidation states and chemical surrounding.
XPS analysis was performed in a VG instrument with a CLAM2
analyser and a twin Mg/Al anode. The pressure in the analysis chamber was
approximately 9⋅10–10 mbar. The measurements were carried out with
unmonochromated AlKα photons (1486.6 eV). The resolution was
determined from the FWHM of the Au (4f7/2) line, which was 1.3 eV with
pass energy of 50 eV. The power of the x-ray source was kept constant at
300 W. The binding energy scale of the spectra was aligned through the C
(1s) peak at 285 or 284.5 eV. Measurements were made using photoelectron
take-off angles of 30° with respect to the surface normal of the sample. The
VGX900 data analysis software was used to calculate the elemental
composition from the peak areas and to analyse the peak positions. Curve
fitting is done using the XPSPEAK version 4.1 program.
B
4.5 Magnetic Measurements
Some of the basic principles of magnetism in solids are discussed in Section
1.5.3 above. In this work the magnetic properties of nanoparticles were
analyzed in a QuantumDesign Physical Properties Measurement System
(PPMS) by using about 15 mg of nanopowder per scan, a magnetic field
range of −6⋅104 to 6⋅104 Oe and with a sensitivity of 2⋅10–5 emu for DC
measurements. The temperature dependence of magnetization was
measured in an applied magnetic field of 50 Oe between 2 and 300 K, using
zero-field-cooling (ZFC) and field-cooling (FC) procedures. The thin films
were analyzed with a QuantumDesign MPMS XL system in the temperature
interval 2-350 K, using magnetic fields (H) from −2⋅104 to 2⋅104 Oe.
4.6 Magnetic Resonance Imaging (MRI)
Since one of the main objectives of the present thesis is to study the
synthesis of novel nanoparticulate contrast enhancing agents for magnetic
resonance imaging (MRI), it is motivated to say a few words about the basic
32
Characterization Techniques
principles behind this important technique [83]. Magnetic resonance
tomography is, theoretically and practically, a highly sophisticated technique,
which is why a more detailed description is beyond the scope of this work.
MRI is a technique for non-invasive imaging of the inside of the human
body, using the principles of nuclear magnetic resonance (NMR). As
hydrogen is the most common atom in living organisms, present in high
concentration in the body’s water as well as in fatty tissues, the 1H nucleus
thus is very appropriate in creating images of the interior of living organisms.
Most atoms possess nuclear spin (I), except when the mass number and
atomic number of a nucleus both happen to be even. The spin makes the
nucleus behave like a small magnetic dipole, and, according to quantum
mechanics, when an external magnetic field is applied to such a nucleus, the
magnetic dipole vector can orient itself in 2I + 1 different ways. As the 1H
nucleus has the spin I = 12 , its magnetic moment can adopt only two
orientations. Either it can align itself with the magnetic field, corresponding
to the low energy state (α), or opposed to the applied field, corresponding to
the high energy state (β), see Fig. 4.5.
β
ΔE
Field off
α
Field on
Figure 4.5. The energy levels of a nucleus with I =
1
2
.
The energy difference between the two states is given by
Δ E = Eβ − Eα = 21 γhB 0 − (− 21 γhB 0 ) = γhB0
(4.6)
where γ is the magnetogyric ratio (42.58 MHz/T for 1H), h is the Planck
constant, and B0 is the magnetic field strength applied. The resonance
frequency, ν, corresponding to a spin transition from the α to β state is
related to the energy difference ΔE by
B
ΔE = hν
(4.7)
Combining Eqs 4.6 and 4.7 and solving for ν gives the resonance frequency
33
Characterization Techniques
ν = γ B0
(4.8)
which is called the Larmor frequency. If the magnetic field strength is kept
constant, different 1H nuclei will resonate at different frequencies depending
on their varying chemical surroundings, which forms the basic principle for
NMR spectroscopy.
The population of the two states is governed by the Boltzmann
distribution equation
N β/N α = e − ΔE/kT
(4.9)
Here Nα and Nβ are the number of spins in the lower and upper levels,
respectively, k is the Boltzmann constant, and T is the absolute temperature.
In an applied magnetic field, B0, Nα will be slightly larger than Nβ,
corresponding to a net magnetization vector, M0, proportional to (Nα − Nβ).
Since ΔE/kT in general is a small number, the quotient Nα/Nβ will be close
to unity. For example, for a magnetic field of 1.5 T at 37 °C Nα/Nβ is only
1.00001. The energy required to excite the 1H nucleus from the lower to the
upper state, corresponds to that of radio frequency waves (RF). After some
time, the excited nucleus will return to its ground state. The relaxation
occurs by two interrelated but different processes, T1 and T2.
When the longitudinal magnetization Mz (where the z axis is taken as
the direction of the external magnetic field, B0) is equal to M0, the system is
at equilibrium. If the system is perturbed by an energy corresponding to that
of the Larmor frequency for a certain lapse of time, the spin system will be
saturated and Mz becomes zero. One can imagine the process so that the Mz
vector experiences a rotation through 90° into the xy plane. When Mz
returns to equilibrium, a time constant called the spin lattice relaxation time (T1)
is used to describe the recovery. The T1 relaxation process is expressed by
the equation
B
B
(
M z = M 0 1 − e −t/T1
)
(4.10)
where t is the time. If in Eq. 4.10 t = T1 then Mz ≈ 0.63M0, as depicted in
Fig. 4.6.
34
Characterization Techniques
Mz
M0
M0(1-e-1)
t
T1
Figure 4.6. The relaxation of Mz to M0.
Initially, all flipped spins will be rotating (precessing) about the z direction
with more or less the same frequency, corresponding to an overall
precession rate of the system in the xy plane equal to the Larmor frequency.
However, as each spin feels a slightly different magnetic field, the individual
frequencies will begin to dephase. Hence, in the T2 process the individual
spin vectors in the xy plane are randomized, and the relaxation of the overall
spin occurs exponentially, as described by the equation
M xy = M xy 0 e −t/T2
(4.11)
where T2 is spin-spin relaxation time. It holds that T2 is always ≤ T1. The decay
of the transverse magnetization, Mxy, is illustrated in Fig. 4.7.
Mxy
Mxy0e-1
t
T2
Figure 4.7. The exponential decay of Mxy.
There are two time constants contributing to the transverse magnetization
decay, so the actual time constant is written as T2* and is called the effective
transverse relaxation time. The two constants arise from molecular interactions,
and magnetic field inhomogenities, and are combined in the equation
35
Characterization Techniques
1/T2* = 1/T2 + 1/T2inhomo
(4.12)
The basic principle behind all magnetic resonance imaging is that the object
to be visualized is exposed to a magnetic field gradient, e.g. along the x
direction. Such a (linear) gradient, Gx, will change the Larmor frequency of
the 1H nuclei along the x axis, according to the equation
ν = γ(B0 + xGx ) = ν 0 + γxGx
⇔
x=
(4.13)
ν − ν0
γG x
In Eq. 4.13 ν0 is the Larmor frequency at the origin of the coordinate system
chosen. In principle, the NMR signal for all (x,y) coordinates in a crosssection of an object can be recorded (for example by successively varying
the direction of the applied gradient) in this way. After subtracting the
background signal, a projection (image) of the cross-section can be
constructed. In practice, of course, the method is a great deal more
complicated, and to create the tomographic image, three types of magnetic
field gradient pulses are applied together with the RF pulse b .
The amplitude, or signal intensity S, of the MR signal depends on
both the pulse sequence used and the T1 and T2 relaxation times. In this
work, a pulse sequence called spin-echo has been used in the spin relaxation
studies. For the spin-echo sequence, the MR signal intensity S is related to
T1 and T2 by the expression
S = k ⋅ ρ (1 − exp(− TR/T1 ))exp(− TE/T2 )
(4.14)
In Eq. 4.14 TR and TE are, respectively, the (pulse) repetition and echo
times, k is a proportionality constant, and ρ is the spin density. (See Papers III
and IV for more details). The MR imaging technique works because T1, T2
and ρ are tissue (and pathology) specific quantities. It is clear from Eq. 4.14
that for a spin-echo pulse sequence, a small T1 value increases the signal
while the opposite is true for a small T2 value.
In all imaging techniques, the visualization of an object depends on
the contrast between the object and the surrounding ones. In MRI the
b
The three magnetic gradients used for the pulses are named slice selective gradient, GS;
phase encoding gradient, Gφ; and frequency encoding gradient, Gf.
36
Characterization Techniques
contrast is defined as the difference in signal intensity between to adjacent
objects, 1 and 2, i.e.
Contrast = S1 − S2
Clearly, if the contrast is poor, a pathological tissue might be difficult to
distinguish from, say, a nearby healthy one. To improve contrast, various
contrast agents are used in MR examinations of patients. Among the most
common contrast agents are paramagnetic chelates, e.g. Gd-DTPA, and
superparamagnetic magnetite (Fe3O4) nanocrystals. When iron oxide
nanocrystals are used, they are usually coated with carbohydrate molecules
in order to improve their bio-compatibility. The presence of a contrast agent
perturbs the 1H spin behaviour by shortening the T1 and/or T2 relaxation
times, thus increasing (positive contrast) or decreasing (negative contrast)
the signal intensity of the studied tissue. The Gd3+ ion is usually regarded as
a positive contrast agent (“T1 agent”), while iron oxide is regarded as a
negative one (“T2 agent”). However, all contrast agents affect to some extent
both the T1 and T2 relaxation times, and the positive or negative character of
a contrast agent also depends on the pulse sequence applied. The relaxivity (ri)
for a contrast agent is defined as
1/Ti = 1/Ti0 + ri ⋅ c , (i = 1, 2)
(4.15)
In Eq. 4.15 Ti is the observed spin lattice (i = 1), or spin-spin (i = 2)
relaxation time, Ti0 is the relaxation time without contrast agent, and c is the
concentration of the contrast agent. It is seen in Eq. 4.15, that the relaxivity
ri is the slope of the (linear) 1/Ti vs c curve, implying that the shorter the Ti
relaxation time, the lager ri value.
In this work the relaxation behaviour induced by nanocrystals of
Gd2O3 and GdFeO3 were studied and compared with that of the
conventional Gd-DTPA chelate. Relaxivity measurements were performed
with a 1.5 T Philips Achieva whole body system. Tubes containing 1 ml of
sample solutions were immersed in a basin containing saline solution kept at
19.5 ± 1.5 °C, and inserted in a head coil. A 2D mixed spin-echo (SE)
sequence interleaved with an inversion recovery (IR) sequence was used
[84,85]. The sets of parameters used were, for shorter Ti times: TE = 30 ms,
TR(SE) = 500 ms, TR(IR) = 1150 ms and IR delay = 150 ms, and for longer
Ti times: TE = 50 ms, TR(SE) = 760 ms, TR(IR) = 2290 ms and IR
delay = 370 ms. From both sets of experiments 1/T1 and 1/T2 results
showing the lowest standard deviation were selected to calculate r1 and r2.
37
Characterization Techniques
4.7 Quantum Chemical Calculations
At the heart of quantum chemistry lies the well known equation by Erwin
Schrödinger (1887-1961) [86]
Hˆ Ψ = EΨ
(4.16)
This equation is valid when the Hamiltonian, Ĥ, is time-independent. Ĥ
operates on Ψ, the wave function, the eigenfunction corresponding to the
eigenvalue E, which is the total energy of the system. Ĥ operates for both
the kinetic and the potential energy of the system. In molecular calculations,
the kinetic energy part is the kinetic energy operators of the electrons and
the potential energy part is the coulombic electrostatic energy of the
electrons and nuclei. The Schrödinger equation is solved with respect to Ψ
and E. The properties of the system can be calculated when Ψ has been
found. In practice, in molecular calculations many approximations need to
be introduced.
In this work the quantum chemical calculations have been performed
using the B3LYP functional [87-89] in the Gaussian03 program [90], (Paper
II).
38
Results and Discussion
5. RESULTS AND DISCUSSION
5.1 Nanoparticle Synthesis (Papers I-IV)
The polyol method (described in Chapter 3.1) yielded yellowish transparent
colloidal dispersions of nanocrystals for the RE2O3, (RE = Gd, Dy, Y)
systems, while in the synthesis of GdFeO3 dark brown precipitates of
nanocrystalline powders were obtained. To promote precipitation of the
RE2O3 particles, various capping molecules were added to the hot reaction
solution, e.g. oleic or citric acid. These molecules chemisorbs via the
carboxylic group to the surface of the oxide particles, forming a
hydrophobic surface layer (see below). The method resulted in a white
precipitate, or brownish syrup. After repeated washing with methanol, offwhite nanocrystalline powders were obtained. Attempts to prepare Gd2O3
nanocrystals using methanol as solvent, and TOPO as capping molecule,
proved unsuccessful.
5.1.1 XRD
The powder XRD patterns of the as-synthesized nanocrystals usually
contained only one very broad, diffuse peak. As mentioned above, this is
due to poor fulfilment of the Bragg condition for extremely small crystallites.
In Fig. 5.1 the powder diffractogram of nanocrystalline Gd2O3 is shown.
The strongest reflection (222) at about 2θ ≈ 28.6° is seen as a broad peak
with low intensity.
For Gd2O3 nanocrystals capped with 16-hydroxyhexadecanoic acid
(HHDA), an additional diffraction peak appeared at 2θ ≈ 21°, (see Fig. 5.2.)
This peak can be ascribed neither to Gd2O3 nor Gd(OH)3. A possible
explanation is that the peak reflects a (semi-)crystalline ordering of the
surface molecular layer, corresponding to an interplanar spacing of ~4.2 Å.
Tandon et al. [91], who studied phase transitions in melt-crystallized oleic
acid, reported that above 12 °C the molten hydrocarbon chains give rise to a
broad diffraction peak centred at d ≈ 4.5 Å. The issue is discussed in more
detail in Paper I.
39
Intensity (arb. units)
Results and Discussion
20
30
40
50
2θ (deg.)
Intensity (arb. units)
Figure 5.1. A typical powder diffractogram of small nanocrystals (< 5 nm).
20
30
40
50
2θ (deg.)
Figure 5.2. Powder diffractogram of HHDA-capped Gd2O3.
When oleic acid was added at the end of the polyol synthesis of GdFeO3,
and after heating the precipitate at 800 °C for 3 h, the XRD pattern revealed
three crystalline phases. Besides GdFeO3 and C-type Gd2O3, a hitherto
unobserved phase appeared, the composition of which is believed to be
Gd3FeO6. In Fig. 5.3, the powder pattern of the sample is shown, where the
novel phase is marked with “+”. Gd3FeO6 is assumed to be isostructural
with orthorhombic Gd3GaO6 (Cmc21 (No. 36), a = 8.99 Å, b = 11.28 Å,
c = 5.48 Å, Z = 4) [92].
40
Results and Discussion
○
● Gd2O3
Intensity (arb. units)
○ GdFeO3
+ Gd3FeO6
(221)
+
○
●
(111)
○
○
○
○
(311)
○+
+
●
○
○●
(312) (042)
○
++
(350)
○
(530)
○
+
+
●
○ ○
15
20
25
30
35
40
45
50
55
○○
60
2θ (deg.)
Figure 5.3. X-ray powder pattern of a sample containing the novel “Gd3FeO6” phase.
The lower curve is the calculated pattern for Gd3FeO6.
In Gd3FeO6, Fe3+ is tetrahedrally coordinated, whereas gadolinium occurs in
two different seven-coordinated crystallographic sites. The proposed
structure of Gd3FeO6 is shown in Fig. 5.4.
Figure 5.4. Proposed structure of Gd3FeO6, isostructural with Gd3GaO6.
Red = O, grey = Gd, turquoise = Fe containing tetrahedra.
41
Results and Discussion
Attempts to synthesize the Gd3FeO6 phase by conventional solid state
methods resulted only in perovskite GdFeO3 plus monoclinic, B-type Gd2O3.
However, when gallium was used instead of iron, Gd3GaO6 was easily
obtained. It was further noted that up to about 50 % of Fe could be
substituted for Ga in Gd3GaO6 without structural changes [93]. In Fig 5.5,
the XRD powder patterns are shown for four different x values of the
composition Gd3Fe1–xGaxO6.
Intensity (arb. units)
x=0
x=0.25
x=0.5
x=1
20
30
40
50
2θ (deg.)
Figure 5.5. X-ray powder patterns of Gd3Fe1–xGaxO6, with x = 1, 0.5, 0.25 and 0.
The lowest curve is the calculated diffractogram for Gd3GaO6.
As iron is incorporated into the structure, the strong (221) peak at 30.20° is
slightly shifted toward lower 2θ values (30.16° for x = 0.5 and 30.10° for
x = 0.25), probably reflecting the slightly larger ionic radius (~3 %) of Fe3+
as compared with Ga3+. For x = 0.25 the (112) peak of GdFeO3 at 32.85°
appears as well as several peaks from monoclinic Gd2O3, and for x = 0 the
two phases are the dominating ones. This leads to the speculation that the
Gd3FeO6 phase is kinetically stabilized and only observable under the special
conditions of colloidal synthesis, as discussed in Section 1.4.4. Since the
novel phase apparently could not be obtained as a pure sample, no full
structural characterization can be presented here.
42
Results and Discussion
Table 5.1. Elemental composition of Gd3Fe1–xGaxO6
for x = 0.5, as measured by EDX.
Element
Gd
Fe
Ga
O
Quantity
3.07 ± 0.4
0.50 ± 0.08
0.46 ± 0.07
6.04 ± 0.5
In Table 5.1, the elemental composition of Gd3Fe1–xGaxO6 with x = 0.5 is
shown, as measured by EDX. The standard deviation is calculated from
three different crystals in the sample. Some grains containing only Gd and O
were also found, implying that small amounts of Gd2O3 were present too.
5.1.2 TEM
The colloidal dispersions of nanocrystals were examined by TEM to
determine the size, shape, and crystallinity of the particles. In Fig. 5.6, a
collection of small (~5 nm) Dy2O3 nanocrystals is shown. The particles are
to some extent aggregated. Similar appearances were also found for Gd2O3
and Y2O3.
Figure 5.6. Dy2O3 nanocrystals.
A TEM micrograph and an SAED pattern of DEG coated perovskite
GdFeO3 nanocrystals are shown in Fig. 5.7. As seen in the figure, the
43
Results and Discussion
particles in this sample are very small with a mean diameter of ~3.5 nm, and
highly crystalline.
(210)
(303) (111)
(212)
(042)
(310)
Figure 5.7. TEM micrograph of GdFeO3 nanocrystals and an SAED pattern.
5.1.3 IR and Raman
As discussed in Chapter 4.3, carboxylic acids chemisorbed onto the surfaces
of metal oxide nanocrystals may bind in different coordination modes.
Depending on the coordination type, the IR bands of the carboxylic group
will be shifted. For example, the single ν(C=O) stretch at 1711 cm–1 for pure
oleic acid splits into two stretching bands, the antisymmetric, νas(COO−), and
symmetric, νs(COO−), bands at 1557 cm–1 and 1443 cm–1, respectively. The
observed IR bands for the organic acids used as capping molecules in this
work are summarized in Table 5.2. The observed bands are in good
agreement with those reported in other studies of carboxylic acids adsorbed
onto metal, or metal oxide, surfaces [94-98]. In this work oleic acid was used
as a capping molecule on Gd2O3, GdFeO3 and Y2O3. The wavenumber
separation, Δ, between the νas(COO−) and νs(COO−) bands reveals that the
coordination of oleic acid, as well as of HHDA and AHDA, are bidentate
and symmetrical. However, the value of the symmetric carboxyl stretch
νs(COO−) is somewhat difficult to interpret due to overlap with the
methylene scissoring band, δ(CH2). For citric and formic acid, the Δ values
indicate, respectively, a bridging and monodentate coordination. The value
for formic acid is quite close to that of the bridging coordination, and the
quantum chemical calculations too point towards the bridge mode as that
with the highest binding energy.
44
Results and Discussion
Table 5.2. Vibrational mode assignments of molecules chemisorbed onto RE2O3
(RE = Gd, Y) and REFeO3 (RE = Gd) nanocrystals (cm–1).
Modes of
vibration
Oleic acid
HHDA1
AHDA2
ν(=C−H)
3006
-
-
-
-
νas(CH3)
2953 (sh)
-
-
-
-
νas(CH2)
2923
2918
2918
x
-
νs(CH2)
2853
2850
2849
x
-
νas(COO−)
1557
1558
1560
1577
1592
δ(HCO)
-
-
-
-
1457
νs(COO−)
1443
1444
1446
1420
1366
δ(CH2)
o
o
o
x
-
ν(C−C)
921
923
921
919
-
ρr(CH2)
722
721
721
x
-
Δ
114
114
114
157
−
Citric acid Formic acid
226
−
−
sh = shoulder, x = not detected, o = overlap with νs(COO ), Δ = νas(COO )−νs(COO ).
1HHDA = 16-hydroxyhexadecanoic acid, 2AHDA = 16-aminohexadecanoic acid.
Thistlethwaite et al. [99] observed that the olefinic CH=CH group of oleic
acid interacts with the surface of rutile TiO2 at pH ≈ 3, resulting in an
absence of the expected ν(=C−H) stretch at 3007 cm–1. In our samples, the
olefinic stretch is visible at 3006 cm–1 (Table 5.2), implying that the double
bond does not interact with the nanoparticle surface. An explanation for this
difference could be that TiIV, which is smaller and has higher charge density
(i.e., is a stronger Lewis acid) than the REIII cations, has the stronger
tendency to interact with the π electrons in the CH=CH group. Lanthanides,
on the other hand, prefer σ bonds and form only very weak bonds with π
ligands [100]. In the same study by Thistlewaite et al., dimerization of the
oleic acid was observed at pH ≈ 3 and 9, and the dimer was adsorbed onto
the TiO2 surface, giving a peak at 1714 cm–1 assigned to the ν(C=O) stretch.
No corresponding observations were made by us for nanocrystalline RE2O3
and GdFeO3 coated with oleic acid.
45
Results and Discussion
The oleic acid capped Gd2O3 particles were also examined with Raman
spectroscopy. The assignments of Raman bands for pure oleic acid, and
oleic acid on Gd2O3, are summarized in Table 5.3. The differences between
the Raman bands of the two samples are in general very small. According to
Tandon et al. [91], the ν(C−C) stretch at 1083 cm–1 should shift to higher
frequencies if the hydrocarbon chains are ordered, a result somewhat at
odds with the interpretation of the XRD and IR results.
Table 5.3. Assignments of Raman bands of pure oleic acid
and oleic acid attached to Gd2O3 nanoparticles (cm–1).
Oleic acid
Gd2O3-OA
ν(=C−H)
3008
3008
νas(CH2)
2897
2895
νs(CH2)
2852
2852
ν(C=C)
1657
1654
δ(CH2)
1439
1439
ρt(CH2)
1303
1302
δ(C=C−H)
1268
1271
ν(CC)
1083
1083
Assignments
5.1.4 XPS
The x-ray photoelectron spectrum (XPS) of the Gd (3d) core level shows a
spin-orbit coupled doublet, corresponding to the 3d5/2 and 3d7/2 electrons.
As seen in Table 5.4, the observed energies for the doublet are identical for
the oleic and citric acid capped particles, and in agreement with previously
published data on Gd2O3 powders [101]. The XPS data thus corroborate
that the particles consist of Gd2O3.
Also listed in Table 5.4 are the O (1s) and C (1s) values for oleic acid
capped Gd2O3 nanocrystals, together with the O (1s) values for the citric
acid capped ones. The O (1s) peak at 531.5 eV for both samples
corresponds to the oxygen in Gd2O3, and the peaks at ~533 eV are assigned
to the carboxylic oxygens in the capping molecules, together with C−O−C
and C−OH oxygens in DEG.
46
Results and Discussion
Table 5.4. XPS binding energies for OA and CA coated Gd2O3 (eV).
Level
Gd2O3-OA
Gd2O3-CA
Gd (3d5/2)
1187.9
1187.9
Gd (3d7/2)
1220.4
1220.4
O (1s)
531.5
531.5
~533
533.6
C (1s)
285
287
289.1
For the oleic acid coated sample also the C (1s) energies were measured. The
peaks at 285 eV and 289.1 eV are both assigned to carbons in oleic acid,
while the peak at 287 eV is more likely attributable to terminal C−OH
carbons in DEG. This, together with the IR data which show bands from
DEG in the region 1123-1047 cm–1, indicate that small amounts of DEG is
still present on the particle surface.
5.1.5 MR
Prior to the MR relaxivity measurements, the samples were filtered and
dialyzed, as described in Papers III and IV. The procedures were carried out
in order to eliminate large particles (or particle agglomerates), and to reduce
the content of free metal ions in the solution. The relaxivity properties of
the nanocrystalline samples (Gd2O3 and GdFeO3) were compared with
those of a commercial gadolinium containing contrast agent, Magnevist®.
Chemically this contrast agent consists of one Gd3+ ion chelated by the
organic molecule diethylenetriaminepentaacetic acid (DTPA). DTPA
coordinates to gadolinium by eight ligand atoms (3 N and 5 O), leaving one
position open for water to coordinate. The complex is very stable, which is
regarded as necessary as gadolinium is to some extent toxic. The relaxivity
properties of the nanocrystals were also compared with those of free Gd3+
ions in deionized water. The results are presented in Table 5.5.
47
Results and Discussion
It is a reasonable presumption that the relaxivity of magnetic nanocrystals
depends not only on the chemical composition of the particles, but on
particle size as well [102]. Since small nanoparticles in contrast to larger ones
have a more favourable surface to volume ratio, and a larger total surface
area, they ought to have higher MR contrast efficacy. If the relaxivity
quotient r2/r1 is low (< 1.5) small paramagnetic nanoparticles may offer
better signal-to-noise properties, and higher resolution in T1-weighted
images, than for example conventional contrast agents.
Table 5.5. Summary of the relaxivity measurements.
r1
r2
Size
(nm)
Dialysis
time (hours)
(mM–1s–1)
(mM–1s–1)
r2/r1
Gd2O3 (G1)
3
96
1.6
3.2
2.1
Gd2O3 (G2)
3
144
6.4
15.2
2.4
Gd2O3-PEG-silane (G3)
3
96
9.4
13.4
1.4
3.5
120
11.9
15.2
1.3
5
4
4.5
11.3
2.5
Gd-DTPA (Magnevist )
-
-
4.1
4.7
1.1
GdCl3
-
-
10.5
12.4
1.2
Sample
GdFeO3 (GF1)
GdFeO3 (GF2)
®
It is seen in Table 5.5 that, in particular, the Gd2O3 sample denoted G3, and
the GdFeO3 sample denoted GF1, showed considerably better r1 relaxivity
values than the reference compound (Gd-DTPA). The difference between
the G1 and G3 samples is that the G3 particles were coated with
polyethylene glycol (PEG) to which a short silane molecule had been
attached, while the G1 particles were coated with DEG. The GF1 sample
showed an r1 value comparable to that of free Gd3+ ions (the only difference
between GF1 and GF2 was that a slightly higher OH–/M ratio was used in
the synthesis of the latter compound; 1.5 instead of 1). For only one of the
samples (G1), was a lower r1 value measured than for the reference. It is
interesting to note that the GF1 sample exhibited a relaxivity behaviour
comparable to that of free Gd3+ ions in water. The material therefore seems
well worth further study.
48
Results and Discussion
5.1.6 Magnetic Properties
In Table 5.6, the saturation magnetization values (emu⋅g−1) of
nanocrystalline Gd2O3 and GdFeO3 at low temperatures (2-5 K) are shown.
The as-synthesized small Gd2O3 (G4) and GdFeO3 (GF3) particles have
saturation values of about the same magnitude. The heat-treated GF3
sample (= GF4), which consisted of considerably larger particles, showed a
somewhat higher saturation value. The magnetization vs field curves for the
particles (not shown here) did not exhibit any hysteresis for G4 gadolinia,
and only very weak hysteresis was observed for the GF4 perovskite sample.
The latter observation seems fairly consistent with the work by Sivakumar et
al. [25], who reported a weak hysteresis for ca 60 nm sized GdFeO3 particles,
prepared by a sonochemical method.
Table 5.6. The saturation magnetization
of some nanocrystalline samples.
Size (nm)
T (K)
Msat (emu⋅g–1)
Gd2O3 (G4)
3
5
70
GdFeO3 (GF3)
3
2
80
GdFeO3 (GF4)*
40
2
135
Sample
* GF3 heated in air at 800 °C for 3 h.
The magnetization vs temperature curves for nanocrystalline Gd2O3 and
GdFeO3 are shown in Figs 5.8a and b. The magnetic behaviour of both
systems is close to paramagnetic. No significant differences between the
zero field cooled (ZFC) and field cooled (FC) susceptibility curves are
observed, indicating that neither Gd2O3 nor GdFeO3 is superparamagnetic.
In recent years, the possibility of superparamagnetism in
nanocrystalline Gd2O3 has been discussed, but so far our magnetic
measurements do not support such a presumption. Interestingly, a split
between the FC and ZFC curves is observed for thin films of GdFeO3
below 5 K, as reported in Paper VI and discussed below.
49
Results and Discussion
0,35
0,3
-1
M (emu·g )
0,2
-1
M (emu·g )
0,25
0,15
0,1
0,05
0,25
0,2
0,15
0,1
0,05
0
0
a)
10
20
30
40
0
50
0
b)
T (K)
10
20
30
40
50
T (K)
Figure 5.8. ZFC (●) and FC (○) curves at H = 50 Oe, for
nanocrystalline a) Gd2O3 (G4) and b) GdFeO3 (GF3).
5.1.7 Quantum Chemical Calculations
Quantum chemical calculations were performed in order to study how
organic molecules, e.g. formic acid and diethylene glycol (DEG,) chemisorb
on the surface of RE2O3 nanoparticles. The results of the calculations are
reported in Paper II. Although we were most interested in studying gadolinia
particles, an yttria cluster (Y12O18) was used as a model system. The reason
for that is that yttria has no unpaired electrons (it is diamagnetic) which
greatly simplifies the quantum chemical calculations. Otherwise Gd2O3 and
Y2O3 are structurally and chemically very similar as discussed in Section
1.4.1. For formic acid, the calculations indicated that it coordinates to the
RE2O3 surface in a bridging or bidentate mode. For the DEG molecule,
calculations were performed for both dissociated and undissociated protons
of the terminal hydroxyl groups. The results of the computations indicated
the most stable adsorption mode to be that of the deprotonated molecule,
with an adsorption energy of 547 kJ⋅mol–1, as compared to the undissociated
value of 228 kJ⋅mol–1.
In Fig. 5.9, the calculated vibration spectra for a dissociated and an
undissociated DEG molecule on an Y12O18 cluster are shown together with
the experimentally collected IR spectrum for DEG capped Y2O3
nanocrystals. The calculated frequencies are somewhat shifted compared to
the ones observed. In spite of this, the calculated spectrum for the
deprotonated DEG molecule shows the best agreement with the measured
50
Results and Discussion
Transmittance (arb. units)
spectrum. A picture of the Y12O18 cluster with the dissociated DEG
molecule is inserted in the spectrum in Fig. 5.9.
1200
1150
1100
1050
1000
950
900
850
Wavenumber (cm−1)
Figure 5.9. a) Calculated IR spectra for a DEG molecule on an Y12O18 cluster. Solid line:
dissociated, and dotted line: undissociated hydroxyl groups. b) Measured IR spectrum of
DEG chemisorbed onto Y2O3 nanoparticles. Inserted: the yttria cluster with one
deprotonated DEG molecule.
51
Results and Discussion
5.2 Thin Film Synthesis (Papers V and VI)
5.2.1 Na0.5K0.5NbO3 (NKN)
Dozens of synthesis procedures were tested in order to prepare NKN thin
films, but most of them resulted in unstable sols in which precipitation
occurred, or failed to properly wet the substrates. The three synthesis routes
described in Paper V are here denoted NK1, NK2 and NK3. The reactants
used are listed in Table 5.7.
Table 5.7. Synthesis routes for NKN thin films.
No.
Solvent
K+
Na+
NbV
Chelating
agents
Other
additives
NK1
Ethanol, Acetic acid
KOAc
NaOAc
NbCl5
AcAc, CA, EG
NH3 ⇒ pH ≈ 4
NK2
Ethanol
K(s)
Na(s)
Nb(OEt)5
-
-
NbCl5
Oxalic acid
-
2-methoxy ethanol,
NK3 Ethanol, Ethylene glycol KOAc NaOAc
In Fig. 5.10, the x-ray powder diffractogram of bulk NKN is shown, and
indexed with the orthorhombic unit cell given by Singh et al. [11], with
a = 3.994 Å, b = 4.016 Å, c = 3.935 Å.
Intensity (arb. units)
(101)
(001)
(020)
(112)
(022)
(002) (012)
(122) (310)
(013)
(111)
15
20
25
30
35
40
45
50
55
60
65
70
75
2θ (deg.)
Figure 5.10. X-ray powder pattern of bulk NKN.
52
80
Results and Discussion
The powder XRD patterns of thin films obtained in the NK1 synthesis
route is shown in Fig. 5.11. The films were made by a modified Pechini
method, as described in Chapter 3.2. The films were heated at three
different temperatures for 1 h, and the peaks indexed by (001), (101) and
(002) are assigned to the NKN phase. The peaks get narrower with
increasing temperatures, indicating grain growth and/or better crystallinity.
For the film annealed at 900 ºC, two extra peaks are observed below 22º
attributable to the (100) and (002) peaks of SiO2, since Si is easily oxidized at
such high temperatures.
(001)
Intensity (arb. units)
(101)
(002)
700°C
800°C
900°C
20
30
40
50
2θ (deg.)
Figure 5.11. X-ray diffractograms of sample NK1 heated at 700-900 °C, for 1 h.
A tentative estimate of the grain sizes of the films from the Scherrer formula
(Eq. 4.2) indicate that the sizes were much smaller than actually observed in
SEM. A possible explanation for this is that the symmetry is not cubic
and/or not single phase, so the peaks seen in the diffractograms do not
correspond to single peaks, but to two or more overlapping peaks with
almost the same 2θ value. SEM images showed the NKN films to consist of
nanosized grains in the range 40-300 nm, mostly of irregular shapes. The
thickness of the films depends on the number of deposited layers, and most
films prepared were about 400 nm (Table 5.8).
In Fig. 5.12, the SEM image of a cross-section of an NK1 film after
heat-treatment at 900 °C is shown. In a study of SrTiO3 thin films, made by
a polymeric precursor method, Pontes et al. [103] observed very sharp
interfaces between each deposited layer of a film calcined at 600 °C. No
such interfaces can be observed between the four deposited layers in Fig.
5.12, though a similar heating process was used.
In brief, the optimal synthesis conditions for well crystallized,
single-phase NKN films are not easily found.
53
Results and Discussion
Table 5.8. Grain-sizes and thicknesses of NKN films, estimated by SEM.
Sample
NK1
NK2
NK3
T (°C)
Estimated
grain-size (nm)
Film thickness
(nm)
700
50-150
-
800
100-200
-
900
100-200
400
600
~200
-
700
100-300
-
800
> 100
500
900
100
-
700
~200
400
800
100-300
-
Figure 5.12. SEM image of a cross-section of NK1, heated at 900 °C.
5.2.2 GdFeO3 & Gd3Fe5O12
The GdFeO3 and Gd3Fe5O12 thin films were, in similarity with the NKN
films, polycrystalline with nanosized grains. The films discussed in Paper VI
are here denoted GF5 and GF6. The XRPD data show the formation of the
expected phases at 700 °C with minor impurities of other phases, which can
be seen in Figs 5.13a and b. For the GdFeO3 sample (GF5), the diffraction
54
Results and Discussion
peaks are slightly shifted towards higher 2θ values compared with the bulk
material. Possibly due to epitaxial strain, the volume of the unit cell
decreases and the diffraction pattern can be fitted by the unit cell of the high
pressure (~2 GPa) phase of GdFeO3 [104]. Thus, the cell parameters of the
orthorhombic cell decrease from a = 5.349 Å, b = 5.611 Å and c = 7.669 Å
for the bulk material, to a = 5.333 Å, b = 5.592 Å and c = 7.643 Å for the
thin film.
Si
X
Intensity (arb. units)
Intensity (arb. units)
Si
X
X
X
a)
20
30
40
50
60
2θ (deg.)
b)
20
30
40
50
2θ (deg.)
Figure 5.13. X-ray powder patterns of thin films of
a) GdFeO3 (GF5) and b) Gd3Fe5O12 (GF6).
For the Gd3Fe5O12 sample (GF6), the diffraction peaks are fitted with the
bulk diffraction pattern, i.e. no change of the unit cell volume was observed.
When studied by SEM, the cross-sections of the films look somewhat
porous. The EDX measurements indicated the films to contain slightly
more iron (between 2 and 5 percent) than expected from stoichiometry. It is
not clear if this extra iron is real, or due to some systematic augmentation of
Fe vis-à-vis Gd in the EDX data. The measured ratios of Gd and Fe are,
anyhow, close to the ones expected (see Table 5.10).
For perovskite GdFeO3, the iron is present only in octahedral sites
and shows only two IR bands. The peaks at 559 cm–1 and 430 cm–1 observed
for the GdFeO3 thin film are assigned to, respectively, the ν(Fe−O)
stretching and δ(O−Fe−O) bending modes of the FeO6 octahedron. The
positions of the bands agree well with those for bulk GdFeO3, as reported
by Sivakumar et al. [25]. Three bands are seen for the Gd3Fe5O12 thin film
and assigned to the antisymmetric iron-oxygen stretches of the FeO4
tetrahedra in the structure [105]. The observed vibration bands for the films
are listed in Table 5.9, and compared to what has been found in the
literature.
55
60
Results and Discussion
Table 5.9. IR vibrational modes of GF5 and GF6 (cm–1).
GdFeO3 (GF5)
Gd3Fe5O12 (GF6)
Our study
[25]
Our study
[106]
-
-
639
638
-
-
585
590
559
560
554
555
430
438
x
435
From the XPS measurements of GF5 and GF6, the binding energy levels of
Gd (3d), Fe (2p) and O (1s) showed the surface composition and oxidation
states of the elements. As shown in Table 5.10, the gadolinium to oxygen
ratios are in good agreement with those expected from the stoichiometry of
both films. However, the iron to oxygen and iron to gadolinium ratios differ
considerably between experimental and expected values. Thus, the iron
content at the surface of the films is only about half of that expected. The
following explanation for this iron deficit is suggested. Since XPS is a
surface sensitive technique, the measured signal from the atoms in the
outmost surface layers are stronger than from those deeper down in the film.
This possibly indicates a crystal growth preferably O−Gd terminated, i.e.
ended with crystal planes more rich in oxygen and gadolinium compared to
iron. If that is true, it is reasonable the measured iron content appears to be
low.
Table 5.10. The experimental and stoichiometric relative ratios of the films.
Gd/O
Fe/O
Gd/Fe
Calc. XPS
Calc. XPS
Calc. XPS EDX
GdFeO3 (GF5)
0.33
0.30
0.33
0.15
1
1.90
0.95
Gd3Fe5O12 (GF6)
0.25
0.24
0.42
0.27
0.6
0.9
0.59
In Table 5.11 the binding energy peak positions of Gd (3d5/2), Fe (2p3/2) and
O (1s) are listed. The peak at 1186.5 eV for Gd (3d5/2) in both films is
56
Results and Discussion
slightly shifted from that of Gd2O3 nanocrystals (Tables 5.4 and 5.11). The
2p3/2 peak of iron at 710.5 eV corresponds to the Fe3+ oxidation state and
shows a small shift compared to pure iron oxide (Fe2O3), the corresponding
peak of which is found at 710.4 eV [107]. The Fe3+ satellite is clearly visible
in both spectra. The satellite appears as a broad peak at around 718 eV,
verifying the +3 oxidation state of iron in the GdFeO3 and Gd3Fe5O12 films.
For both films the O (1s) binding energy position at 529 eV is attributed to
lattice oxygens, and is consistent with XPS results on GdFeO3 powders
published earlier [108].
Table 5.11. The XPS binding energy positions of GF5 and GF6 (eV).
Gd (3d5/2)
Fe (2p3/2)
O (1s) (lattice)
GdFeO3 (GF5)
1186.5
710.5
529
Gd3Fe5O12 (GF6)
1186.5
710.5
529
The magnetization vs temperature curve for the GdFeO3 and Gd3Fe5O12
thin films are shown in Figs 5.14a and b, respectively. The magnetization of
GdFeO3 can be described as the sum of two contributing terms; one due to
the spontaneous magnetic ordering of the iron containing sublattice, and
one susceptibility term mainly attributable to the paramagnetic gadolinium
sublattice. This is mathematically expressed by the equation
M(T) = M0(T) + χ(T) ⋅ H
where M is the magnetization, M0 is a small ferromagnetic moment of the
basically antiferromagnetic iron containing sublattice, χ is the magnetic
susceptibility, and T is the temperature.
Below ~5 K, the FC and ZFC curves for GdFeO3 exhibit a split,
indicative of either an antiferromagnetic coupling between the gadolinium
and iron sublattices, or a spin-glass behaviour [109]. The garnet phase is, as
expected, ferrimagnetic and shows a compensation temperature
Tcomp ≈ 295 K.
Giess and Potemski [110] showed that the compensation temperature
of Gd3Fe5O12 thin films is not fixed but can be varied by varying the Gd:Fe
ratio and the calcination temperature. In this way, the compensation
temperature could be varied between 277 K and 291 K. The same
observation was made by Oron et al. [111]. They found that for Gd3Fe5O12
57
Results and Discussion
films heat-treated at 750 °C, 900 °C and 1100 °C, Tcomp values of,
respectively, 278 K, 286 K and 298 K were obtained.
70
7
6
60
-1
M (emu·g )
6
4
50
M (emu·g-1)
-1
M (emu·g )
5
5
40
4
0
5
T (K)
10
3
30
2
20
1
10
0
a)
0
0
50
100
150
T (K)
200
250
300
b)
0
50
100
150
200
250
300
T (K)
Figure 5.14. ZFC (●) and FC (○) curves at H = 1000 Oe, for thin films of
a) GdFeO3 (GF5) and b) Gd3Fe5O12 (GF6).
The behaviour of the magnetization vs field curves for nanocrystalline
GdFeO3 might be interpreted as size-dependent, as shown in Fig. 5.15. For
small particles (GF3, < 5 nm) the magnetization saturates at ~80 emu⋅g–1,
while for larger particles and thin films (GF4, ~40 nm, and GF5, ~300 nm),
the behaviour is almost linear at lower fields. However, the magnetization
for the larger nanocrystals (GF4) saturates at about 135 emu⋅g–1 at much
higher fields. Sivakumar et al. [25] investigated the magnetic properties of
nanocrystalline GdFeO3 with an average diameter of about 60 nm. They
found a similar “linear” M/H dependence, and a weak hysteresis, for the
material.
58
350
Results and Discussion
80
60
40
M (emu·g-1)
20
0
-25000 -20000 -15000 -10000
-5000
0
5000
10000
15000
20000
25000
-20
-40
-60
-80
H (Oe)
Figure 5.15. M vs H curves at 2 K, for GdFeO3 nanocrystals and thin film.
GF3 (+), GF4 (●) and GF5 (○).
59
Results and Discussion
60
Conclusions
6. CONCLUSIONS
It has been demonstrated that a variety of metal oxide nanocrystals and thin
films can be prepared by colloidal chemistry. Highly crystalline sub-15 nm
particles of RE2O3 (RE = Y, Gd, Dy) and GdFeO3 have been synthesized
with a polyol method, using diethylene glycol both as solvent and as capping
molecule. At temperatures around 180-210 °C, nanocrystals of just a few
nanometres in diameter were formed. The nanocrystals can be coated
(functionalized) with carboxylic acids, as evidenced by FT-IR and XPS. It
has been shown by both FT-IR and quantum chemical calculations that the
carboxylic groups bind to the metal ions on the particle surface in a
bidentate or bridging mode. In samples where GdFeO3 particles were
precipitated with oleic acid and heated at 800 °C, a hitherto unknown phase
was observed. The suggested composition of the phase is Gd3FeO6 and is
believed to be isostructural with Gd3GaO6. The result suggests that certain
phases not observable in the bulk state, nonetheless can be kinetically
stabilized in the nano state. Attempts to synthesize Gd3FeO6 with
conventional solid state methods only yielded the thermodynamically stable
phases GdFeO3 and Gd2O3. However, it was shown that Fe3+ can be
incorporated into bulk Gd3GaO6 with up to about 50 % without structural
changes.
Nanocrystalline samples of Gd2O3 and GdFeO3 were examined with
respect to magnetic resonance relaxivity and magnetic susceptibility. Both
materials show promising properties for the use as positive contrast
enhancing agents in MRI, with relaxivities higher than the reference Gd3+
complex. Further studies with regard to functionalization, stability and
biocompatibility are needed to demonstrate the viability of the proposed
materials.
Polycrystalline thin films of Na0.5K0.5NbO3, GdFeO3 and Gd3Fe5O12
were prepared with various sol-gel techniques. At temperatures as low as
< 600 °C, crystallization of the films was observed by XRD. Although
minor amounts of impurities were found in the films, the chances seem
good that single phase materials can be prepared. FT-IR measurements were
61
Conclusions
consistent with the presence of the GdFeO3 and Gd3Fe5O12 phases in the
films.
Magnetic measurements in the temperature range 2-350 K indicated
that the magnetization of GdFeO3 can be described as the sum of two
contributing terms. One term (mainly) due to the spontaneous magnetic
ordering of the iron-containing sublattice, and the other a susceptibility term,
attributable to the paramagnetic gadolinium sublattice. The two terms yield
the relationship M(T) = M0(T) + χ(T) ⋅ H for the magnetization. Below
about 5 K, the FC and ZFC curves for GdFeO3 showed a split, consistent
with either an antiferromagnetic, or a spin-glass behaviour. The garnet phase,
on the other hand, is ferrimagnetic and showed a compensation temperature
Tcomp ≈ 295 K. The XPS measurements corroborated the +3 oxidation state
of iron. However, they also indicated a lower iron concentration than
expected at the surface of the films. A possible explanation for this iron
“deficit” is that the crystal growth preferably ends in crystal planes rich in
gadolinium and oxygen but not in iron.
62
Acknowledgments
ACKNOWLEDGMENTS
“Thank you for the voice, thank you for the eyes, thank you for the good times”
Simple Minds, 1981.
Vi är alla lite som atomer. Vi interagerar med andra, bildar vackra molekyler
och underbara strukturer som berikar tillvaron. Det här arbetet hade inte
blivit vad det är utan de människor jag haft omkring mig under min tid som
doktorand här i Linköping. Jag vill därför tacka alla stort, som på något sätt
bidragit till att jag nått detta mål.
Tack till…
…Per-Olov Käll, min handledare som alltid funnits till hands och hjälpt
mig med mångt och mycket, med sin kunskap och uppmuntran. Tack för
alla intressanta diskussioner vi haft om kemi, konst, litteratur m.m.
…alla i fysikalisk/oorganisk kemi som jag lärt känna. Lars O, Misha, NilsOla, Claes-Göran, Richard, Annika, Maria L (tack Maria för att du läste
igenom mitt manuskript så bra). Tack även till ex-jobbare och före detta
medarbetare.
…Kajsa Uvdal, min biträdande handledare, och Bo Liedberg (och
Stiftelsen för Strategisk Forskning, SSF Grant No. SSF[A3 05:204]) för att
jag fick den här möjligheten att genomföra detta.
…Rita Fantl och Susanne Andersson, som varit mig behjälpliga med
allehanda pappersarbete.
…alla jag jobbat med på kemiavdelningen. Ni är många och ni vet vilka ni är.
…Henrik och Rodrigo, som arbetat i projektet, men som även berikat
tiden utanför arbetet.
…David Lawrence för språkgranskningen av min avhandling.
…alla doktorandkollegor för trevliga stunder: Lan, Marcus, Timmy, Leffe,
Roger, Patrik, Johan, Alma och Andreas, m.fl. som funnits här genom
63
Acknowledgments
åren. Och självklart ett stort tack till The Coffee Break Girls: Karin A, Cissi,
Ina, Anngelica, Sara, Sofie, Veronica, Karin S, Patricia, (vad hade
arbetsdagarna varit utan er?).
…Maria A, Linnéa, Maria E, Anna K, Marc, för det trevliga samarbetet i
projektet. Tack även till Daniel A, Stina B, Fredrik B, Cissi V med flera
vid Tillämpad fysik.
…personer jag jobbat med vid Tunnfilmsfysik och Plasma &
Beläggningsfysik, speciellt då Ulf Helmersson tillsammans med Veronika
K och Denis M för de första åren av min tid som doktorand. Tack även till
Jens B för all hjälp med diffraktometrar och annat. Många tack till Per,
Axel och David för all värdefull hjälp jag fått med TEM. Stort tack till
Thomas Lingefelt som alltid funnits till hands och varit behjälplig med
elektronmikroskopen och annan utrustning.
…Louise för alla fina samtal vi haft.
…Xiaodong Zou m.fl. vid Strukturkemi i Stockholm.
…min familj som alltid stöttat mig, och tack till alla fina vänner jag har
utanför universitetsvärlden, ni är fantastiska!
64
References
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
C.B. Murray, S. Sun, W. Gaschler, H. Doyle, T.A. Betley, C.R. Kagan, IBM J. Res.
& Dev. 45 (2001) 47-56.
C.N.R. Rao, G.U. Kulkarni, P.J. Thomas, P.P. Edwards, Chem. Eur. J. 8 (2002)
28-35.
The Royal Society & the Royal Academy of Engineering (2004) Nanoscience and
nanotechnologies: opportunities and uncertainties. London.
U. Helmersson, S. Todorova, S.A. Barnett, J.E. Sundgren, L.C. Markert, J.E.
Greene, J. Appl. Phys. 62 (1987) 481-484.
G. Sberveglieri, Sens. Actuators B 23 (1995) 103-109.
K.K. Shung, J.M. Cannata, Q.F. Zhou, J. Electroceram. 19 (2007) 139-145.
R.P. Vollertsen, Microelectron. Reliab. 43 (2003) 865-878.
A.H. Soloway, W. Tjarks, B.A. Barnum, F.G. Rong, R.F. Barth, I.M. Codogni,
J.G. Wilson, Chem. Rev. 98 (1998) 1515-1562.
J.L. Bridot, A.C. Faure, S. Laurent, C. Rivière, C. Billotey, B. Hiba, M. Janier, V.
Josserand, J.L. Coll, L.V. Elst, R. Muller, S. Roux, P. Perriat, O. Tillement, J. Am.
Chem. Soc. 129 (2007) 5076-5084.
http://en.wikipedia.org/wiki/perovskite
K. Singh, V. Lingwal, S.C. Bhatt, N.S. Panwar, B.S. Semwal, Mater. Res. Bull. 36
(2001) 2365-2374.
V.J. Tennery, K.W. Hang, J. Appl. Phys. 39 (1968) 4749-4753.
A.C. Sakowski-Cowley, K. Lukaszewicz, H.D. Megaw, Acta Cryst. B 25 (1969)
851-865.
A.W. Hewat, J. Phys. C 6 (1973) 2559-2572.
L.E. Cross, Nature 181 (1958) 178-179.
R.E. Jaeger, L. Egerton. J. Am. Ceram. Soc. 45 (1962) 209-213.
R.H. Dungan, R.D. Golding. J. Am. Ceram. Soc. 48 (1965) 601.
L. Egerton, D.M. Dillon, J. Am. Ceram. Soc. 42 (1959) 438-442.
S. Lanfredi, L. Dessemond, A.C. Martins Rodrigues, J. Eur. Ceram. Soc. 20 (2000)
983-990.
G. Shirane, R. Newnham, R. Pepinsky, Phys. Rev. 96 (1954) 581-588.
S. Geller, E.A. Wood, Acta Cryst. 9 (1956) 563-568.
I. Dzyaloshinsky, J. Phys. Chem. Solids 4 (1958) 241-255.
O. Nikolov, I. Hall, S.N. Barilo, A.A. Mukhin, J. Magn. Magn. Mater. 152 (1996)
75-85.
X.S. Niu, W.M. Du, W.P. Du, Sens. Actuators B 99 (2004) 399-404.
M. Sivakumar, A. Gedanken, D. Bhattacharya, I. Brukental, Y. Yeshurun, W.
Zhong, Y.W. Du, I. Felner, I. Nowik, Chem. Mater. 16 (2004) 3623-3632.
S. Mathur, H. Shen, N. Lecerf, A. Kjekshus, H. Fjellvåg, G.F. Goya, Adv. Mater.
14 (2002) 1405-1409.
S.V. Chavan, A.K. Tyagi, J. Mater. Res. 20 (2005) 2654-2659.
F. Bertaut, F. Forrat, A. Herpin, P. Mériel, Compt. Rend. 243 (1956) 898-901.
65
References
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
[55]
[56]
[57]
[58]
[59]
[60]
[61]
M. Sugimoto, J. Am. Ceram. Soc. 82 (1999) 269-280.
F. Bertaut, F. Forrat, Compt. Rend. 242 (1956) 382-384.
J.E. Weidenborner, Acta Cryst. 14 (1961) 1051-1056.
L. Néel, Compt. Rend. 239 (1954) 8-11.
I. Warshaw, R. Roy, J. Am. Ceram. Soc. 42 (1959) 434-438.
A.N. Thakur, K. Gaur, H.B. Lal, J. Mater. Sci. Lett. 11 (1992) 496-497.
T. Matsumoto, N. Môri, J. Iida, M. Tanaka, K. Siratori, F. Izumi, H. Asano,
Physica B 180/181 (1992) 603-605.
J. Nicolas, J. Coutures, J.P. Coutures, B. Boudot, J. Solid State Chem. 52 (1984)
101-113.
J.C. Guitel, M. Marezio, J. Mareschal, Mater. Res. Bull. 11 (1976) 739-744.
M. Inoue, T. Nishikawa, T. Nakamura, T. Inui, J. Am. Ceram. Soc. 80 (1997) 21572160.
Y. Mizoguchi, H. Onodera, H. Yamauchi, M. Kagawa, Y. Syono, T. Hirai, Mater.
Sci. Eng. A 217/218 (1996) 164-166.
A.A. Bossak, I.E. Graboy, O.Y. Gorbenko, A.R. Kaul, M.S. Kartavtseva, V.L.
Svetchnikov, H.W. Zandbergen, Chem. Mater. 16 (2004) 1751-1755.
W. Ostwald, Z. Phys. Chem. 22 (1897) 289-330.
T. Threlfall, Org. Process Res. Dev. 7 (2003) 1017-1027.
P.O. Käll, IFM, Linköping University, personal communication.
M. Faraday, Philos. Trans. R. Soc. London 147 (1857) 145-181.
D. Thompson, Gold Bull. 40 (2007) 267-269.
W. Ostwald. 1896. Lehrbruck der Allgemeinen Chemie, vol. 2, part 1. Leipzig,
Germany.
K.J. Klabunde, D. Zhang, G.N. Glavee, C.M. Sorensen, G.C. Hadjipanayis, Chem.
Mater. 6 (1994) 784-787.
X.M. Lin, C.M. Sorensen, K.J. Klabunde, G.C. Hadjipanayis, Langmuir 14 (1998)
7140-7146.
D.H. Chen, S.H. Wu, Chem. Mater. 12 (2000) 1354-1360.
J.A. Nelson, L.H. Bennett, M.J. Wagner, J. Am. Chem. Soc. 124 (2002) 2979-2983.
N.A.D. Burke, H.D.H. Stöver, F.P. Dawson, J.D. Lavers, P.K. Jain, H. Oka,
IEEE Trans. Magn. 37 (2001) 2660-2662.
D.P. Dinega, M.G. Bawendi, Angew. Chem. Int. Ed. Engl. 38 (1999) 1788-1791.
M.L. Steigerwald, A.P. Alivisatos, J.M. Gibson, T.D. Harris, R. Kortan, A.J.
Muller, A.M. Thayer, T.M. Duncan, D.C. Douglass, L.E. Brus, J. Am. Chem. Soc.
110 (1988) 3046-3050.
M.L. Steigerwald, L.E. Brus, Annu. Rev. Mater. Sci. 19 (1989) 471-495.
L. Motte, F. Billoudet, M.P. Pileni, J. Phys. Chem. 99 (1995) 16425-16429.
C.B. Murray, C.R. Kagan, M.G. Bawendi, Science 270 (1995) 1335-1338.
M.P. Pileni, J. Phys. Chem. B 105 (2001) 3358-3371.
M.P. Pileni, J. Phys. Condens. Matter 18 (2006) S67-S84.
E.V. Shevchenko, D.V. Talapin, N.A. Kotov, S. O’Brien, C.B. Murray, Nature
439 (2006) 55-59.
D.V. Talapin, E.V. Shevchenko, C.B. Murray, A.V. Titov, P. Král, Nano Lett. 7
(2007) 1213-1219.
G. Wakefield, H.A. Keron, P.J. Dobson, J.L. Hutchison, J. Colloid Interface Sci. 215
(1999) 179-182.
66
References
[62]
[63]
[64]
[65]
[66]
[67]
[68]
[69]
[70]
[71]
[72]
[73]
[74]
[75]
[76]
[77]
[78]
[79]
[80]
[81]
[82]
[83]
[84]
[85]
[86]
[87]
[88]
[89]
[90]
[91]
[92]
[93]
[94]
[95]
[96]
G. Wakefield, H.A. Keron, P.J. Dobson, J.L. Hutchison, J. Phys. Chem. Solids 60
(1999) 503-508.
R. Bazzi, M.A. Flores-Gonzalez, C. Louis, K. Lebbou, C. Dujardin, A. Brenier,
W. Zhang, O. Tillement, E. Bernstein, P. Perriat, J. Lumin. 102/103 (2003) 445450.
F. Fievet, J.P. Lagier, B. Blin, B. Beaudoin, M. Figlarz, Solid State Ionics 32/33
(1989) 198-205.
C. Feldmann, Adv. Funct. Mater. 13 (2003) 101-107.
D. Caruntu, Y. Remond, N.H. Chou, M.J. Jun, G. Caruntu, J. He, G. Goloverda,
C. O’Connor, V. Kolesnichenko, Inorg. Chem. 41 (2002) 6137-6146.
A.K. Boal, K. Das, M. Gray, V.M. Rotello, Chem. Mater. 14 (2002) 2628-2636.
A. Dierstein, H. Natter, F. Meyer, H.O. Stephan, C. Kropf, R. Hempelmann,
Scripta Mater. 44 (2001) 2209-2212.
M.T. Reetz, W. Helbig, J. Am. Chem. Soc. 116 (1994) 7401-7402.
H. Natter, R. Hempelmann, Electrochim. Acta 49 (2003) 51-61.
W.W. Zhang, W.P. Zhang, P.B. Xie, M. Yin, H.T. Chen, L. Jing, Y.S. Zhang, L.R.
Lou, S.D. Xia, J. Colloid Interface Sci. 262 (2003) 588-593.
D.M. Roy, R. Roy, Am. Mineral. 39 (1954) 957-975.
R. Roy, J. Am. Ceram. Soc. 39 (1956) 145-146.
R. Roy, J. Am. Ceram. Soc. 52 (1969) 344.
J.J. Ebelmen, Ann. Chimie Phys. 16 (1846) 129-166.
J.J. Ebelmen, Compt. Rend. 25 (1847) 854-856.
T. Graham, J. Chem. Soc. 17 (1864) 318-327.
M.P. Pechini, US Pat. 3,330,697 (1967).
C.L. Mak, B. Lai, K.H. Wong, C.L. Choy, D. Mo, Y.L. Zhang, J. Appl. Phys. 89
(2001) 4491-4496.
C. Baratto, P.P. Lottici, D. Bersani, G. Antonioli, G. Gnappi, A. Montenero,
J. Sol-Gel Sci. Technol. 13 (1998) 667-671.
K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination Compounds, 5th
ed.; John Wiley & Sons: New York, (1997); Part B, p. 59.
M. Nara, H. Torii, M. Tasumi, J. Phys. Chem. 100 (1996) 19812-19817.
J.P. Hornak, The Basics of MRI, http://www.cis.rit.edu/htbooks/mri/
M. Engström, A. Klasson, H. Pedersen, C. Vahlberg, P.O. Käll, K. Uvdal, Magn.
Reson. Mater. Phys. 19 (2006) 180-186.
J.J. In den Kleef, J.J. Cuppen, Magn. Reson. Med. 5 (1987) 513-524.
E. Schrödinger, Phys. Rev. 28 (1926) 1049-1070.
A.D. Becke, J. Chem. Phys. 98 (1993) 1372-1377.
A.D. Becke, J. Chem. Phys. 98 (1993) 5648-5652.
C. Lee, W. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785-789.
Gaussian03, Revision B.05, M.J. Frisch et al. Gaussian Inc. Wallinford, CT, 2003.
P. Tandon, G. Förster, R. Neubert, S. Wartewig, J. Mol. Struct. 524 (2000) 201-215.
H. Yamane, T. Sakamoto, S.I. Kubota, M. Shimada, Acta Cryst. C 55 (1999) 479481.
F. Söderlind et al., to be published.
Y.T. Tao, J. Am. Chem. Soc. 115 (1993) 4350-4358.
S. Pawsey, K. Yach, J. Halla, L. Reven, Langmuir 16 (2000) 3294-3303.
N. Shukla, C. Liu, P.M. Jones, D. Weller, J. Magn. Magn. Mater. 266 (2003) 178184.
67
References
[97]
[98]
[99]
[100]
[101]
[102]
[103]
[104]
[105]
[106]
[107]
[108]
[109]
[110]
[111]
N. Wu, L. Fu, M. Su, M. Aslam, K.C. Wong, V.P. Dravid, Nano Lett. 4 (2004)
383-386.
H.J. Choi, S.W. Han, S.J. Lee, K. Kim, J. Colloid Interface Sci. 264 (2003) 458-466.
P.J. Thistlethwaite, M.S. Hook, Langmuir. 16 (2000) 4993-4998.
N.N. Greenwood, A. Earnshaw, Chemistry of the Elements, 2nd Ed. ButterworthHeinemann, Oxford (1997) p. 1238, and N. Kaltsoyannis, P. Scott, The f elements,
Oxford University Press, Oxford (1999) p. 69.
D. Raiser, J.P. Deville, J. Electron Spectrosc. Relat. Phenom. 57 (1991) 91-97.
Y.W. Jun, Y.M. Huh, J.S. Choi, J.H. Lee, H.T. Song, S. Kim, S. Yoon, K.S. Kim,
J.S. Shin, J.S. Suh, J. Cheon, J. Am. Chem. Soc. 127 (2005) 5732-5733.
F.M. Pontes, E.R. Leite, E.J.H. Lee, E. Longo, J.A. Varela, J. Eur. Ceram. Soc. 21
(2001) 419-426.
N.L. Ross, J. Zhao, R.J. Angel, J. Solid State Chem. 177 (2004) 3768-3775.
A.M. Hofmeister, K.R. Campbell, J. Appl. Phys. 72 (1992) 638-646.
N.T. McDevitt, J. Opt. Soc. Am. 59 (1969) 1240-1244.
P.C.J. Graat, M.A.J. Somers, Appl. Surf. Sci. 100/101 (1996) 36-40.
H. Aono, E. Traversa, M. Sakamoto, Y. Sadaoka, Sens. Actuators B 94 (2003) 132139.
Y. Ogo, H. Yanagi, T. Kamiya, K. Nomura, M. Hirano, H. Hosono, J. Appl. Phys.
101 (2007) 103714.
E.A. Giess, R.M. Potemski, IBM Tech. Disclosure Bull. 10 (1967) 852.
M. Oron, I. Barlow, W.F. Traber, J. Mater. Sci. 4 (1969) 271-281.
68